I don't have anything else queued up at the moment, so this is just Richard's SME patches. -- PMM The following changes since commit 63b38f6c85acd312c2cab68554abf33adf4ee2b3: Merge tag 'pull-target-arm-20220707' of https://git.linaro.org/people/pmaydell/qemu-arm into staging (2022-07-08 06:17:11 +0530) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20220711 for you to fetch changes up to f9982ceaf26df27d15547a3a7990a95019e9e3a8: linux-user/aarch64: Add SME related hwcap entries (2022-07-11 13:43:52 +0100) ---------------------------------------------------------------- target-arm: * Implement SME emulation, for both system and linux-user ---------------------------------------------------------------- Richard Henderson (45): target/arm: Handle SME in aarch64_cpu_dump_state target/arm: Add infrastructure for disas_sme target/arm: Trap non-streaming usage when Streaming SVE is active target/arm: Mark ADR as non-streaming target/arm: Mark RDFFR, WRFFR, SETFFR as non-streaming target/arm: Mark BDEP, BEXT, BGRP, COMPACT, FEXPA, FTSSEL as non-streaming target/arm: Mark PMULL, FMMLA as non-streaming target/arm: Mark FTSMUL, FTMAD, FADDA as non-streaming target/arm: Mark SMMLA, UMMLA, USMMLA as non-streaming target/arm: Mark string/histo/crypto as non-streaming target/arm: Mark gather/scatter load/store as non-streaming target/arm: Mark gather prefetch as non-streaming target/arm: Mark LDFF1 and LDNF1 as non-streaming target/arm: Mark LD1RO as non-streaming target/arm: Add SME enablement checks target/arm: Handle SME in sve_access_check target/arm: Implement SME RDSVL, ADDSVL, ADDSPL target/arm: Implement SME ZERO target/arm: Implement SME MOVA target/arm: Implement SME LD1, ST1 target/arm: Export unpredicated ld/st from translate-sve.c target/arm: Implement SME LDR, STR target/arm: Implement SME ADDHA, ADDVA target/arm: Implement FMOPA, FMOPS (non-widening) target/arm: Implement BFMOPA, BFMOPS target/arm: Implement FMOPA, FMOPS (widening) target/arm: Implement SME integer outer product target/arm: Implement PSEL target/arm: Implement REVD target/arm: Implement SCLAMP, UCLAMP target/arm: Reset streaming sve state on exception boundaries target/arm: Enable SME for -cpu max linux-user/aarch64: Clear tpidr2_el0 if CLONE_SETTLS linux-user/aarch64: Reset PSTATE.SM on syscalls linux-user/aarch64: Add SM bit to SVE signal context linux-user/aarch64: Tidy target_restore_sigframe error return linux-user/aarch64: Do not allow duplicate or short sve records linux-user/aarch64: Verify extra record lock succeeded linux-user/aarch64: Move sve record checks into restore linux-user/aarch64: Implement SME signal handling linux-user: Rename sve prctls linux-user/aarch64: Implement PR_SME_GET_VL, PR_SME_SET_VL target/arm: Only set ZEN in reset if SVE present target/arm: Enable SME for user-only linux-user/aarch64: Add SME related hwcap entries docs/system/arm/emulation.rst | 4 + linux-user/aarch64/target_cpu.h | 5 +- linux-user/aarch64/target_prctl.h | 62 +- target/arm/cpu.h | 7 + target/arm/helper-sme.h | 126 ++++ target/arm/helper-sve.h | 4 + target/arm/helper.h | 18 + target/arm/translate-a64.h | 45 ++ target/arm/translate.h | 16 + target/arm/sme-fa64.decode | 60 ++ target/arm/sme.decode | 88 +++ target/arm/sve.decode | 41 +- linux-user/aarch64/cpu_loop.c | 9 + linux-user/aarch64/signal.c | 243 ++++++-- linux-user/elfload.c | 20 + linux-user/syscall.c | 28 +- target/arm/cpu.c | 35 +- target/arm/cpu64.c | 11 + target/arm/helper.c | 56 +- target/arm/sme_helper.c | 1140 +++++++++++++++++++++++++++++++++++++ target/arm/sve_helper.c | 28 + target/arm/translate-a64.c | 103 +++- target/arm/translate-sme.c | 373 ++++++++++++ target/arm/translate-sve.c | 393 ++++++++++--- target/arm/translate-vfp.c | 12 + target/arm/translate.c | 2 + target/arm/vec_helper.c | 24 + target/arm/meson.build | 3 + 28 files changed, 2821 insertions(+), 135 deletions(-) create mode 100644 target/arm/sme-fa64.decode create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c From: Richard Henderson <richard.henderson@linaro.org> Dump SVCR, plus use the correct access check for Streaming Mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-2-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) int i; int el = arm_current_el(env); const char *ns_status; + bool sve; qemu_fprintf(f, " PC=%016" PRIx64 " ", env->pc); for (i = 0; i < 32; i++) { @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) el, psr & PSTATE_SP ? 'h' : 't'); + if (cpu_isar_feature(aa64_sme, cpu)) { + qemu_fprintf(f, " SVCR=%08" PRIx64 " %c%c", + env->svcr, + (FIELD_EX64(env->svcr, SVCR, ZA) ? 'Z' : '-'), + (FIELD_EX64(env->svcr, SVCR, SM) ? 'S' : '-')); + } if (cpu_isar_feature(aa64_bti, cpu)) { qemu_fprintf(f, " BTYPE=%d", (psr & PSTATE_BTYPE) >> 10); } @@ -XXX,XX +XXX,XX @@ static void aarch64_cpu_dump_state(CPUState *cs, FILE *f, int flags) qemu_fprintf(f, " FPCR=%08x FPSR=%08x\n", vfp_get_fpcr(env), vfp_get_fpsr(env)); - if (cpu_isar_feature(aa64_sve, cpu) && sve_exception_el(env, el) == 0) { + if (cpu_isar_feature(aa64_sme, cpu) && FIELD_EX64(env->svcr, SVCR, SM)) { + sve = sme_exception_el(env, el) == 0; + } else if (cpu_isar_feature(aa64_sve, cpu)) { + sve = sve_exception_el(env, el) == 0; + } else { + sve = false; + } + + if (sve) { int j, zcr_len = sve_vqm1_for_el(env, el); for (i = 0; i <= FFR_PRED_NUM; i++) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This includes the build rules for the decoder, and the new file for translation, but excludes any instructions. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-3-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 1 + target/arm/sme.decode | 20 ++++++++++++++++++++ target/arm/translate-a64.c | 7 ++++++- target/arm/translate-sme.c | 35 +++++++++++++++++++++++++++++++++++ target/arm/meson.build | 2 ++ 5 files changed, 64 insertions(+), 1 deletion(-) create mode 100644 target/arm/sme.decode create mode 100644 target/arm/translate-sme.c diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) } bool disas_sve(DisasContext *, uint32_t); +bool disas_sme(DisasContext *, uint32_t); void gen_gvec_rax1(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, uint32_t opr_sz, uint32_t max_sz); diff --git a/target/arm/sme.decode b/target/arm/sme.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME instruction descriptions +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } switch (extract32(insn, 25, 4)) { - case 0x0: case 0x1: case 0x3: /* UNALLOCATED */ + case 0x0: + if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { + unallocated_encoding(s); + } + break; + case 0x1: case 0x3: /* UNALLOCATED */ unallocated_encoding(s); break; case 0x2: diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ +/* + * AArch64 SME translation + * + * Copyright (c) 2022 Linaro, Ltd + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library; if not, see <http://www.gnu.org/licenses/>. + */ + +#include "qemu/osdep.h" +#include "cpu.h" +#include "tcg/tcg-op.h" +#include "tcg/tcg-op-gvec.h" +#include "tcg/tcg-gvec-desc.h" +#include "translate.h" +#include "exec/helper-gen.h" +#include "translate-a64.h" +#include "fpu/softfloat.h" + + +/* + * Include the generated decoder. + */ + +#include "decode-sme.c.inc" diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), + decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), @@ -XXX,XX +XXX,XX @@ arm_ss.add(when: 'TARGET_AARCH64', if_true: files( 'sme_helper.c', 'translate-a64.c', 'translate-sve.c', + 'translate-sme.c', )) arm_softmmu_ss = ss.source_set() -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This new behaviour is in the ARM pseudocode function AArch64.CheckFPAdvSIMDEnabled, which applies to AArch32 via AArch32.CheckAdvSIMDOrFPEnabled when the EL to which the trap would be delivered is in AArch64 mode. Given that ARMv9 drops support for AArch32 outside EL0, the trap EL detection ought to be trivially true, but the pseudocode still contains a number of conditions, and QEMU has not yet committed to dropping A32 support for EL[12] when v9 features are present. Since the computation of SME_TRAP_NONSTREAMING is necessarily different for the two modes, we might as well preserve bits within TBFLAG_ANY and allocate separate bits within TBFLAG_A32 and TBFLAG_A64 instead. Note that DDI0616A.a has typos for bits [22:21] of LD1RO in the table of instructions illegal in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-4-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 7 +++ target/arm/translate.h | 4 ++ target/arm/sme-fa64.decode | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/helper.c | 41 +++++++++++++++++ target/arm/translate-a64.c | 40 ++++++++++++++++- target/arm/translate-vfp.c | 12 +++++ target/arm/translate.c | 2 + target/arm/meson.build | 1 + 8 files changed, 195 insertions(+), 2 deletions(-) create mode 100644 target/arm/sme-fa64.decode diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A32, HSTR_ACTIVE, 9, 1) * the same thing as the current security state of the processor! */ FIELD(TBFLAG_A32, NS, 10, 1) +/* + * Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. + * This requires an SME trap from AArch32 mode when using NEON. + */ +FIELD(TBFLAG_A32, SME_TRAP_NONSTREAMING, 11, 1) /* * Bit usage when in AArch32 state, for M-profile only. @@ -XXX,XX +XXX,XX @@ FIELD(TBFLAG_A64, SMEEXC_EL, 20, 2) FIELD(TBFLAG_A64, PSTATE_SM, 22, 1) FIELD(TBFLAG_A64, PSTATE_ZA, 23, 1) FIELD(TBFLAG_A64, SVL, 24, 4) +/* Indicates that SME Streaming mode is active, and SMCR_ELx.FA64 is not. */ +FIELD(TBFLAG_A64, SME_TRAP_NONSTREAMING, 28, 1) /* * Helpers for using the above. diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ typedef struct DisasContext { bool pstate_sm; /* True if PSTATE.ZA is set. */ bool pstate_za; + /* True if non-streaming insns should raise an SME Streaming exception. */ + bool sme_trap_nonstreaming; + /* True if the current instruction is non-streaming. */ + bool is_nonstreaming; /* True if MVE insns are definitely not predicated by VPR or LTPSIZE */ bool mve_no_pred; /* diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 SME allowed instruction decoding +# +# Copyright (c) 2022 Linaro, Ltd +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, see <http://www.gnu.org/licenses/>. + +# +# This file is processed by scripts/decodetree.py +# + +# These patterns are taken from Appendix E1.1 of DDI0616 A.a, +# Arm Architecture Reference Manual Supplement, +# The Scalable Matrix Extension (SME), for Armv9-A + +{ + [ + OK 0-00 1110 0000 0001 0010 11-- ---- ---- # SMOV W|Xd,Vn.B[0] + OK 0-00 1110 0000 0010 0010 11-- ---- ---- # SMOV W|Xd,Vn.H[0] + OK 0100 1110 0000 0100 0010 11-- ---- ---- # SMOV Xd,Vn.S[0] + OK 0000 1110 0000 0001 0011 11-- ---- ---- # UMOV Wd,Vn.B[0] + OK 0000 1110 0000 0010 0011 11-- ---- ---- # UMOV Wd,Vn.H[0] + OK 0000 1110 0000 0100 0011 11-- ---- ---- # UMOV Wd,Vn.S[0] + OK 0100 1110 0000 1000 0011 11-- ---- ---- # UMOV Xd,Vn.D[0] + ] + FAIL 0--0 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD vector operations +} + +{ + [ + OK 0101 1110 --1- ---- 11-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar) + OK 0101 1110 -10- ---- 00-1 11-- ---- ---- # FMULX/FRECPS/FRSQRTS (scalar, FP16) + OK 01-1 1110 1-10 0001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar) + OK 01-1 1110 1111 1001 11-1 10-- ---- ---- # FRECPE/FRSQRTE/FRECPX (scalar, FP16) + ] + FAIL 01-1 111- ---- ---- ---- ---- ---- ---- # Advanced SIMD single-element operations +} + +FAIL 0-00 110- ---- ---- ---- ---- ---- ---- # Advanced SIMD structure load/store +FAIL 1100 1110 ---- ---- ---- ---- ---- ---- # Advanced SIMD cryptography extensions +FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS + +# These are the "avoidance of doubt" final table of Illegal Advanced SIMD instructions +# We don't actually need to include these, as the default is OK. +# -001 111- ---- ---- ---- ---- ---- ---- # Scalar floating-point operations +# --10 110- ---- ---- ---- ---- ---- ---- # Load/store pair of FP registers +# --01 1100 ---- ---- ---- ---- ---- ---- # Load FP register (PC-relative literal) +# --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) +# --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) +# --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) + +FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR +FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA +FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT +FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS +FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR +FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP +FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) +FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA +FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL +FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD +FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA +FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA +FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions +FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) +FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) +FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) +FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) +FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) +FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) +FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) +FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) +FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) +FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) +FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) +FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch +FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) +FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) +FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) +FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ int sme_exception_el(CPUARMState *env, int el) return 0; } +/* This corresponds to the ARM pseudocode function IsFullA64Enabled(). */ +static bool sme_fa64(CPUARMState *env, int el) +{ + if (!cpu_isar_feature(aa64_sme_fa64, env_archcpu(env))) { + return false; + } + + if (el <= 1 && !el_is_in_host(env, el)) { + if (!FIELD_EX64(env->vfp.smcr_el[1], SMCR, FA64)) { + return false; + } + } + if (el <= 2 && arm_is_el2_enabled(env)) { + if (!FIELD_EX64(env->vfp.smcr_el[2], SMCR, FA64)) { + return false; + } + } + if (arm_feature(env, ARM_FEATURE_EL3)) { + if (!FIELD_EX64(env->vfp.smcr_el[3], SMCR, FA64)) { + return false; + } + } + + return true; +} + /* * Given that SVE is enabled, return the vector length for EL. */ @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a32(CPUARMState *env, int fp_el, DP_TBFLAG_ANY(flags, PSTATE__IL, 1); } + /* + * The SME exception we are testing for is raised via + * AArch64.CheckFPAdvSIMDEnabled(), as called from + * AArch32.CheckAdvSIMDOrFPEnabled(). + */ + if (el == 0 + && FIELD_EX64(env->svcr, SVCR, SM) + && (!arm_is_el2_enabled(env) + || (arm_el_is_aa64(env, 2) && !(env->cp15.hcr_el2 & HCR_TGE))) + && arm_el_is_aa64(env, 1) + && !sme_fa64(env, el)) { + DP_TBFLAG_A32(flags, SME_TRAP_NONSTREAMING, 1); + } + return rebuild_hflags_common_32(env, fp_el, mmu_idx, flags); } @@ -XXX,XX +XXX,XX @@ static CPUARMTBFlags rebuild_hflags_a64(CPUARMState *env, int el, int fp_el, } if (FIELD_EX64(env->svcr, SVCR, SM)) { DP_TBFLAG_A64(flags, PSTATE_SM, 1); + DP_TBFLAG_A64(flags, SME_TRAP_NONSTREAMING, !sme_fa64(env, el)); } DP_TBFLAG_A64(flags, PSTATE_ZA, FIELD_EX64(env->svcr, SVCR, ZA)); } diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void do_vec_ld(DisasContext *s, int destidx, int element, * unallocated-encoding checks (otherwise the syndrome information * for the resulting exception will be incorrect). */ -static bool fp_access_check(DisasContext *s) +static bool fp_access_check_only(DisasContext *s) { if (s->fp_excp_el) { assert(!s->fp_access_checked); @@ -XXX,XX +XXX,XX @@ static bool fp_access_check(DisasContext *s) return true; } +static bool fp_access_check(DisasContext *s) +{ + if (!fp_access_check_only(s)) { + return false; + } + if (s->sme_trap_nonstreaming && s->is_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, false)); + return false; + } + return true; +} + /* Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. */ @@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, uint32_t insn, bool isread, default: g_assert_not_reached(); } - if ((ri->type & ARM_CP_FPU) && !fp_access_check(s)) { + if ((ri->type & ARM_CP_FPU) && !fp_access_check_only(s)) { return; } else if ((ri->type & ARM_CP_SVE) && !sve_access_check(s)) { return; @@ -XXX,XX +XXX,XX @@ static void disas_data_proc_simd_fp(DisasContext *s, uint32_t insn) } } +/* + * Include the generated SME FA64 decoder. + */ + +#include "decode-sme-fa64.c.inc" + +static bool trans_OK(DisasContext *s, arg_OK *a) +{ + return true; +} + +static bool trans_FAIL(DisasContext *s, arg_OK *a) +{ + s->is_nonstreaming = true; + return true; +} + /** * is_guarded_page: * @env: The cpu environment @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, dc->mte_active[1] = EX_TBFLAG_A64(tb_flags, MTE0_ACTIVE); dc->pstate_sm = EX_TBFLAG_A64(tb_flags, PSTATE_SM); dc->pstate_za = EX_TBFLAG_A64(tb_flags, PSTATE_ZA); + dc->sme_trap_nonstreaming = EX_TBFLAG_A64(tb_flags, SME_TRAP_NONSTREAMING); dc->vec_len = 0; dc->vec_stride = 0; dc->cp_regs = arm_cpu->cp_regs; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) } } + s->is_nonstreaming = false; + if (s->sme_trap_nonstreaming) { + disas_sme_fa64(s, insn); + } + switch (extract32(insn, 25, 4)) { case 0x0: if (!extract32(insn, 31, 1) || !disas_sme(s, insn)) { diff --git a/target/arm/translate-vfp.c b/target/arm/translate-vfp.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.c +++ b/target/arm/translate-vfp.c @@ -XXX,XX +XXX,XX @@ static bool vfp_access_check_a(DisasContext *s, bool ignore_vfp_enabled) return false; } + /* + * Note that rebuild_hflags_a32 has already accounted for being in EL0 + * and the higher EL in A64 mode, etc. Unlike A64 mode, there do not + * appear to be any insns which touch VFP which are allowed. + */ + if (s->sme_trap_nonstreaming) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_Streaming, + s->base.pc_next - s->pc_curr == 2)); + return false; + } + if (!s->vfp_enabled && !ignore_vfp_enabled) { assert(!arm_dc_feature(s, ARM_FEATURE_M)); unallocated_encoding(s); diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static void arm_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs) dc->vec_len = EX_TBFLAG_A32(tb_flags, VECLEN); dc->vec_stride = EX_TBFLAG_A32(tb_flags, VECSTRIDE); } + dc->sme_trap_nonstreaming = + EX_TBFLAG_A32(tb_flags, SME_TRAP_NONSTREAMING); } dc->cp_regs = cpu->cp_regs; dc->features = env->features; diff --git a/target/arm/meson.build b/target/arm/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/meson.build +++ b/target/arm/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('sve.decode', extra_args: '--decode=disas_sve'), decodetree.process('sme.decode', extra_args: '--decode=disas_sme'), + decodetree.process('sme-fa64.decode', extra_args: '--static-decode=disas_sme_fa64'), decodetree.process('neon-shared.decode', extra_args: '--decode=disas_neon_shared'), decodetree.process('neon-dp.decode', extra_args: '--decode=disas_neon_dp'), decodetree.process('neon-ls.decode', extra_args: '--decode=disas_neon_ls'), -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark ADR as a non-streaming instruction, which should trap if full a64 support is not enabled in streaming mode. Removing entries from sme-fa64.decode is an easy way to see what remains to be done. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-5-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.h | 7 +++++++ target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 8 ++++---- 3 files changed, 11 insertions(+), 5 deletions(-) diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ uint64_t asimd_imm_const(uint32_t imm, int cmode, int op); static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ { return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); } +#define TRANS_FEAT_NONSTREAMING(NAME, FEAT, FUNC, ...) \ + static bool trans_##NAME(DisasContext *s, arg_##NAME *a) \ + { \ + s->is_nonstreaming = true; \ + return dc_isar_feature(FEAT, s) && FUNC(s, __VA_ARGS__); \ + } + #endif /* TARGET_ARM_TRANSLATE_H */ diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1010 ---- ---- ---- # ADR FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_adr(DisasContext *s, arg_rrri *a, gen_helper_gvec_3 *fn) return gen_gvec_ool_zzz(s, fn, a->rd, a->rn, a->rm, a->imm); } -TRANS_FEAT(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) -TRANS_FEAT(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) -TRANS_FEAT(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) -TRANS_FEAT(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) +TRANS_FEAT_NONSTREAMING(ADR_p32, aa64_sve, do_adr, a, gen_helper_sve_adr_p32) +TRANS_FEAT_NONSTREAMING(ADR_p64, aa64_sve, do_adr, a, gen_helper_sve_adr_p64) +TRANS_FEAT_NONSTREAMING(ADR_s32, aa64_sve, do_adr, a, gen_helper_sve_adr_s32) +TRANS_FEAT_NONSTREAMING(ADR_u32, aa64_sve, do_adr, a, gen_helper_sve_adr_u32) /* *** SVE Integer Misc - Unpredicated Group -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-6-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 9 ++++++--- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0010 0101 --01 100- 1111 000- ---0 ---- # RDFFR, RDFFRS -FAIL 0010 0101 --10 1--- 1001 ---- ---- ---- # WRFFR, SETFFR FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_predset(DisasContext *s, int esz, int rd, int pat, bool setflag) TRANS_FEAT(PTRUE, aa64_sve, do_predset, a->esz, a->rd, a->pat, a->s) /* Note pat == 31 is #all, to set all elements. */ -TRANS_FEAT(SETFFR, aa64_sve, do_predset, 0, FFR_PRED_NUM, 31, false) +TRANS_FEAT_NONSTREAMING(SETFFR, aa64_sve, + do_predset, 0, FFR_PRED_NUM, 31, false) /* Note pat == 32 is #unimp, to set no elements. */ TRANS_FEAT(PFALSE, aa64_sve, do_predset, 0, a->rd, 32, false) @@ -XXX,XX +XXX,XX @@ static bool trans_RDFFR_p(DisasContext *s, arg_RDFFR_p *a) .rd = a->rd, .pg = a->pg, .s = a->s, .rn = FFR_PRED_NUM, .rm = FFR_PRED_NUM, }; + + s->is_nonstreaming = true; return trans_AND_pppp(s, &alt_a); } -TRANS_FEAT(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) -TRANS_FEAT(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) +TRANS_FEAT_NONSTREAMING(RDFFR, aa64_sve, do_mov_p, a->rd, FFR_PRED_NUM) +TRANS_FEAT_NONSTREAMING(WRFFR, aa64_sve, do_mov_p, FFR_PRED_NUM, a->rn) static bool do_pfirst_pnext(DisasContext *s, arg_rr_esz *a, void (*gen_fn)(TCGv_i32, TCGv_ptr, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-7-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 22 ++++++++++++---------- 2 files changed, 12 insertions(+), 13 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0000 0100 --1- ---- 1011 -0-- ---- ---- # FTSSEL, FEXPA -FAIL 0000 0101 --10 0001 100- ---- ---- ---- # COMPACT -FAIL 0100 0101 --0- ---- 1011 ---- ---- ---- # BDEP, BEXT, BGRP FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_2 * const fexpa_fns[4] = { NULL, gen_helper_sve_fexpa_h, gen_helper_sve_fexpa_s, gen_helper_sve_fexpa_d, }; -TRANS_FEAT(FEXPA, aa64_sve, gen_gvec_ool_zz, - fexpa_fns[a->esz], a->rd, a->rn, 0) +TRANS_FEAT_NONSTREAMING(FEXPA, aa64_sve, gen_gvec_ool_zz, + fexpa_fns[a->esz], a->rd, a->rn, 0) static gen_helper_gvec_3 * const ftssel_fns[4] = { NULL, gen_helper_sve_ftssel_h, gen_helper_sve_ftssel_s, gen_helper_sve_ftssel_d, }; -TRANS_FEAT(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, ftssel_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(FTSSEL, aa64_sve, gen_gvec_ool_arg_zzz, + ftssel_fns[a->esz], a, 0) /* *** SVE Predicate Logical Operations Group @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(TRN2_q, aa64_sve_f64mm, gen_gvec_ool_arg_zzz, static gen_helper_gvec_3 * const compact_fns[4] = { NULL, NULL, gen_helper_sve_compact_s, gen_helper_sve_compact_d }; -TRANS_FEAT(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, compact_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(COMPACT, aa64_sve, gen_gvec_ool_arg_zpz, + compact_fns[a->esz], a, 0) /* Call the helper that computes the ARM LastActiveElement pseudocode * function, scaled by the element size. This includes the not found @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3 * const bext_fns[4] = { gen_helper_sve2_bext_b, gen_helper_sve2_bext_h, gen_helper_sve2_bext_s, gen_helper_sve2_bext_d, }; -TRANS_FEAT(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bext_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BEXT, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bext_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bdep_fns[4] = { gen_helper_sve2_bdep_b, gen_helper_sve2_bdep_h, gen_helper_sve2_bdep_s, gen_helper_sve2_bdep_d, }; -TRANS_FEAT(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bdep_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BDEP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bdep_fns[a->esz], a, 0) static gen_helper_gvec_3 * const bgrp_fns[4] = { gen_helper_sve2_bgrp_b, gen_helper_sve2_bgrp_h, gen_helper_sve2_bgrp_s, gen_helper_sve2_bgrp_d, }; -TRANS_FEAT(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, - bgrp_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(BGRP, aa64_sve2_bitperm, gen_gvec_ool_arg_zzz, + bgrp_fns[a->esz], a, 0) static gen_helper_gvec_3 * const cadd_fns[4] = { gen_helper_sve2_cadd_b, gen_helper_sve2_cadd_h, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-8-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 24 +++++++++++++++--------- 2 files changed, 15 insertions(+), 11 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 000- ---- 0110 1--- ---- ---- # PMULLB, PMULLT (128b result) -FAIL 0110 0100 --1- ---- 1110 01-- ---- ---- # FMMLA, BFMMLA FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_trans_pmull(DisasContext *s, arg_rrr_esz *a, bool sel) gen_helper_gvec_pmull_q, gen_helper_sve2_pmull_h, NULL, gen_helper_sve2_pmull_d, }; - if (a->esz == 0 - ? !dc_isar_feature(aa64_sve2_pmull128, s) - : !dc_isar_feature(aa64_sve, s)) { + + if (a->esz == 0) { + if (!dc_isar_feature(aa64_sve2_pmull128, s)) { + return false; + } + s->is_nonstreaming = true; + } else if (!dc_isar_feature(aa64_sve, s)) { return false; } return gen_gvec_ool_arg_zzz(s, fns[a->esz], a, sel); @@ -XXX,XX +XXX,XX @@ DO_ZPZZ_FP(FMINP, aa64_sve2, sve2_fminp_zpzz) * SVE Integer Multiply-Add (unpredicated) */ -TRANS_FEAT(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_s, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) -TRANS_FEAT(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, gen_helper_fmmla_d, - a->rd, a->rn, a->rm, a->ra, 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_s, aa64_sve_f32mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_s, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FMMLA_d, aa64_sve_f64mm, gen_gvec_fpst_zzzz, + gen_helper_fmmla_d, a->rd, a->rn, a->rm, a->ra, + 0, FPST_FPCR) static gen_helper_gvec_4 * const sqdmlal_zzzw_fns[] = { NULL, gen_helper_sve2_sqdmlal_zzzw_h, @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, TRANS_FEAT(BFDOT_zzxz, aa64_sve_bf16, gen_gvec_ool_arg_zzxz, gen_helper_gvec_bfdot_idx, a) -TRANS_FEAT(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_bfmmla, a, 0) +TRANS_FEAT_NONSTREAMING(BFMMLA, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_bfmmla, a, 0) static bool do_BFMLAL_zzzw(DisasContext *s, arg_rrrr_esz *a, bool sel) { -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-9-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 15 +++++++++++---- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0110 0101 --0- ---- 0000 11-- ---- ---- # FTSMUL -FAIL 0110 0101 --01 0--- 100- ---- ---- ---- # FTMAD -FAIL 0110 0101 --01 1--- 001- ---- ---- ---- # FADDA FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static gen_helper_gvec_3_ptr * const ftmad_fns[4] = { NULL, gen_helper_sve_ftmad_h, gen_helper_sve_ftmad_s, gen_helper_sve_ftmad_d, }; -TRANS_FEAT(FTMAD, aa64_sve, gen_gvec_fpst_zzz, - ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, - a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) +TRANS_FEAT_NONSTREAMING(FTMAD, aa64_sve, gen_gvec_fpst_zzz, + ftmad_fns[a->esz], a->rd, a->rn, a->rm, a->imm, + a->esz == MO_16 ? FPST_FPCR_F16 : FPST_FPCR) /* *** SVE Floating Point Accumulating Reduction Group @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) if (a->esz == 0 || !dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_FADDA(DisasContext *s, arg_rprr_esz *a) DO_FP3(FADD_zzz, fadd) DO_FP3(FSUB_zzz, fsub) DO_FP3(FMUL_zzz, fmul) -DO_FP3(FTSMUL, ftsmul) DO_FP3(FRECPS, recps) DO_FP3(FRSQRTS, rsqrts) #undef DO_FP3 +static gen_helper_gvec_3_ptr * const ftsmul_fns[4] = { + NULL, gen_helper_gvec_ftsmul_h, + gen_helper_gvec_ftsmul_s, gen_helper_gvec_ftsmul_d +}; +TRANS_FEAT_NONSTREAMING(FTSMUL, aa64_sve, gen_gvec_fpst_arg_zzz, + ftsmul_fns[a->esz], a, 0) + /* *** SVE Floating Point Arithmetic - Predicated Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-10-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 12 ++++++------ 2 files changed, 6 insertions(+), 7 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --0- ---- 1001 10-- ---- ---- # SMMLA, UMMLA, USMMLA FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMLALT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, false, true) TRANS_FEAT(FMLSLB_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, false) TRANS_FEAT(FMLSLT_zzxw, aa64_sve2, do_FMLAL_zzxw, a, true, true) -TRANS_FEAT(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_smmla_b, a, 0) -TRANS_FEAT(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_usmmla_b, a, 0) -TRANS_FEAT(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, - gen_helper_gvec_ummla_b, a, 0) +TRANS_FEAT_NONSTREAMING(SMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_smmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(USMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_usmmla_b, a, 0) +TRANS_FEAT_NONSTREAMING(UMMLA, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, + gen_helper_gvec_ummla_b, a, 0) TRANS_FEAT(BFDOT_zzzz, aa64_sve_bf16, gen_gvec_ool_arg_zzzz, gen_helper_gvec_bfdot, a, 0) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-11-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 1 - target/arm/translate-sve.c | 35 ++++++++++++++++++----------------- 2 files changed, 18 insertions(+), 18 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 0100 0101 --1- ---- 1--- ---- ---- ---- # SVE2 string/histo/crypto instructions FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ DO_SVE2_ZZZ_NARROW(RSUBHNT, rsubhnt) static gen_helper_gvec_flags_4 * const match_fns[4] = { gen_helper_sve2_match_ppzz_b, gen_helper_sve2_match_ppzz_h, NULL, NULL }; -TRANS_FEAT(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(MATCH, aa64_sve2, do_ppzz_flags, a, match_fns[a->esz]) static gen_helper_gvec_flags_4 * const nmatch_fns[4] = { gen_helper_sve2_nmatch_ppzz_b, gen_helper_sve2_nmatch_ppzz_h, NULL, NULL }; -TRANS_FEAT(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) +TRANS_FEAT_NONSTREAMING(NMATCH, aa64_sve2, do_ppzz_flags, a, nmatch_fns[a->esz]) static gen_helper_gvec_4 * const histcnt_fns[4] = { NULL, NULL, gen_helper_sve2_histcnt_s, gen_helper_sve2_histcnt_d }; -TRANS_FEAT(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, - histcnt_fns[a->esz], a, 0) +TRANS_FEAT_NONSTREAMING(HISTCNT, aa64_sve2, gen_gvec_ool_arg_zpzz, + histcnt_fns[a->esz], a, 0) -TRANS_FEAT(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, - a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) +TRANS_FEAT_NONSTREAMING(HISTSEG, aa64_sve2, gen_gvec_ool_arg_zzz, + a->esz == 0 ? gen_helper_sve2_histseg : NULL, a, 0) DO_ZPZZ_FP(FADDP, aa64_sve2, sve2_faddp_zpzz) DO_ZPZZ_FP(FMAXNMP, aa64_sve2, sve2_fmaxnmp_zpzz) @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(SQRDCMLAH_zzzz, aa64_sve2, gen_gvec_ool_zzzz, TRANS_FEAT(USDOT_zzzz, aa64_sve_i8mm, gen_gvec_ool_arg_zzzz, a->esz == 2 ? gen_helper_gvec_usdot_b : NULL, a, 0) -TRANS_FEAT(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, - gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) +TRANS_FEAT_NONSTREAMING(AESMC, aa64_sve2_aes, gen_gvec_ool_zz, + gen_helper_crypto_aesmc, a->rd, a->rd, a->decrypt) -TRANS_FEAT(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, false) -TRANS_FEAT(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, - gen_helper_crypto_aese, a, true) +TRANS_FEAT_NONSTREAMING(AESE, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, false) +TRANS_FEAT_NONSTREAMING(AESD, aa64_sve2_aes, gen_gvec_ool_arg_zzz, + gen_helper_crypto_aese, a, true) -TRANS_FEAT(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4e, a, 0) -TRANS_FEAT(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, - gen_helper_crypto_sm4ekey, a, 0) +TRANS_FEAT_NONSTREAMING(SM4E, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4e, a, 0) +TRANS_FEAT_NONSTREAMING(SM4EKEY, aa64_sve2_sm4, gen_gvec_ool_arg_zzz, + gen_helper_crypto_sm4ekey, a, 0) -TRANS_FEAT(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, gen_gvec_rax1, a) +TRANS_FEAT_NONSTREAMING(RAX1, aa64_sve2_sha3, gen_gvec_fn_arg_zzz, + gen_gvec_rax1, a) TRANS_FEAT(FCVTNT_sh, aa64_sve2, gen_gvec_fpst_arg_zpz, gen_helper_sve2_fcvtnt_sh, a, 0, FPST_FPCR) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-12-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 9 --------- target/arm/translate-sve.c | 6 ++++++ 2 files changed, 6 insertions(+), 9 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 10-- ---- ---- ---- # SVE2 32-bit gather NT load (vector+scalar) FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) -FAIL 1000 010- -01- ---- 1--- ---- ---- ---- # SVE 32-bit gather load (vector+imm) -FAIL 1000 0100 0-0- ---- 0--- ---- ---- ---- # SVE 32-bit gather load byte (scalar+vector) -FAIL 1000 0100 1--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load half (scalar+vector) -FAIL 1000 0101 0--- ---- 0--- ---- ---- ---- # SVE 32-bit gather load word (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch -FAIL 1110 010- -00- ---- 001- ---- ---- ---- # SVE2 64-bit scatter NT store (vector+scalar) -FAIL 1110 010- -10- ---- 001- ---- ---- ---- # SVE2 32-bit scatter NT store (vector+scalar) -FAIL 1110 010- ---- ---- 1-0- ---- ---- ---- # SVE scatter store (scalar+32-bit vector) -FAIL 1110 010- ---- ---- 101- ---- ---- ---- # SVE scatter store (misc) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LD1_zpiz(DisasContext *s, arg_LD1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDNT1_zprz(DisasContext *s, arg_LD1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_ST1_zpiz(DisasContext *s, arg_ST1_zpiz *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STNT1_zprz(DisasContext *s, arg_ST1_zprz *a) if (!dc_isar_feature(aa64_sve2, s)) { return false; } + s->is_nonstreaming = true; if (!sve_access_check(s)) { return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. In this case, introduce PRF_ns (prefetch non-streaming) to handle the checks. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-13-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/sve.decode | 10 +++++----- target/arm/translate-sve.c | 11 +++++++++++ 3 files changed, 16 insertions(+), 8 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1000 010- -00- ---- 111- ---- ---- ---- # SVE 32-bit gather prefetch (vector+imm) -FAIL 1000 0100 0-1- ---- 0--- ---- ---- ---- # SVE 32-bit gather prefetch (scalar+vector) FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) -FAIL 1100 010- ---- ---- ---- ---- ---- ---- # SVE 64-bit gather load/prefetch diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ LD1RO_zpri 1010010 .. 01 0.... 001 ... ..... ..... \ @rpri_load_msz nreg=0 # SVE 32-bit gather prefetch (scalar plus 32-bit scaled offsets) -PRF 1000010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1000010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 32-bit gather prefetch (vector plus immediate) -PRF 1000010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1000010 -- 00 ----- 111 --- ----- 0 ---- # SVE contiguous prefetch (scalar plus immediate) PRF 1000010 11 1- ----- 0-- --- ----- 0 ---- @@ -XXX,XX +XXX,XX @@ LD1_zpiz 1100010 .. 01 ..... 1.. ... ..... ..... \ @rpri_g_load esz=3 # SVE 64-bit gather prefetch (scalar plus 64-bit scaled offsets) -PRF 1100010 00 11 ----- 1-- --- ----- 0 ---- +PRF_ns 1100010 00 11 ----- 1-- --- ----- 0 ---- # SVE 64-bit gather prefetch (scalar plus unpacked 32-bit scaled offsets) -PRF 1100010 00 -1 ----- 0-- --- ----- 0 ---- +PRF_ns 1100010 00 -1 ----- 0-- --- ----- 0 ---- # SVE 64-bit gather prefetch (vector plus immediate) -PRF 1100010 -- 00 ----- 111 --- ----- 0 ---- +PRF_ns 1100010 -- 00 ----- 111 --- ----- 0 ---- ### SVE Memory Store Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PRF_rr(DisasContext *s, arg_PRF_rr *a) return true; } +static bool trans_PRF_ns(DisasContext *s, arg_PRF_ns *a) +{ + if (!dc_isar_feature(aa64_sve, s)) { + return false; + } + /* Prefetch is a nop within QEMU. */ + s->is_nonstreaming = true; + (void)sve_access_check(s); + return true; +} + /* * Move Prefix * -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-14-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 2 -- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) -FAIL 1010 010- ---- ---- 011- ---- ---- ---- # SVE contiguous FF load (scalar+scalar) -FAIL 1010 010- ---1 ---- 101- ---- ---- ---- # SVE contiguous NF load (scalar+imm) FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDFF1_zprr(DisasContext *s, arg_rprr_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LDNF1_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { int vsz = vec_full_reg_size(s); int elements = vsz >> dtype_esz[a->dtype]; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Mark these as a non-streaming instructions, which should trap if full a64 support is not enabled in streaming mode. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-15-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme-fa64.decode | 3 --- target/arm/translate-sve.c | 2 ++ 2 files changed, 2 insertions(+), 3 deletions(-) diff --git a/target/arm/sme-fa64.decode b/target/arm/sme-fa64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme-fa64.decode +++ b/target/arm/sme-fa64.decode @@ -XXX,XX +XXX,XX @@ FAIL 0001 1110 0111 1110 0000 00-- ---- ---- # FJCVTZS # --11 1100 --0- ---- ---- ---- ---- ---- # Load/store FP register (unscaled imm) # --11 1100 --1- ---- ---- ---- ---- --10 # Load/store FP register (register offset) # --11 1101 ---- ---- ---- ---- ---- ---- # Load/store FP register (scaled imm) - -FAIL 1010 010- -01- ---- 000- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+scalar) -FAIL 1010 010- -010 ---- 001- ---- ---- ---- # SVE load & replicate 32 bytes (scalar+imm) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zprr(DisasContext *s, arg_rprr_load *a) if (a->rm == 31) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), dtype_msz(a->dtype)); @@ -XXX,XX +XXX,XX @@ static bool trans_LD1RO_zpri(DisasContext *s, arg_rpri_load *a) if (!dc_isar_feature(aa64_sve_f64mm, s)) { return false; } + s->is_nonstreaming = true; if (sve_access_check(s)) { TCGv_i64 addr = new_tmp_a64(s); tcg_gen_addi_i64(addr, cpu_reg_sp(s, a->rn), a->imm * 32); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These functions will be used to verify that the cpu is in the correct state for a given instruction. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-16-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 21 +++++++++++++++++++++ target/arm/translate-a64.c | 34 ++++++++++++++++++++++++++++++++++ 2 files changed, 55 insertions(+) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void write_fp_dreg(DisasContext *s, int reg, TCGv_i64 v); bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn, unsigned int imms, unsigned int immr); bool sve_access_check(DisasContext *s); +bool sme_enabled_check(DisasContext *s); +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned); + +/* This function corresponds to CheckStreamingSVEEnabled. */ +static inline bool sme_sm_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK); +} + +/* This function corresponds to CheckSMEAndZAEnabled. */ +static inline bool sme_za_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_ZA_MASK); +} + +/* Note that this function corresponds to CheckStreamingSVEAndZAEnabled. */ +static inline bool sme_smza_enabled_check(DisasContext *s) +{ + return sme_enabled_check_with_svcr(s, R_SVCR_SM_MASK | R_SVCR_ZA_MASK); +} + TCGv_i64 clean_data_tbi(DisasContext *s, TCGv_i64 addr); TCGv_i64 gen_mte_check1(DisasContext *s, TCGv_i64 addr, bool is_write, bool tag_checked, int log2_size); diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool sme_access_check(DisasContext *s) return true; } +/* This function corresponds to CheckSMEEnabled. */ +bool sme_enabled_check(DisasContext *s) +{ + /* + * Note that unlike sve_excp_el, we have not constrained sme_excp_el + * to be zero when fp_excp_el has priority. This is because we need + * sme_excp_el by itself for cpregs access checks. + */ + if (!s->fp_excp_el || s->sme_excp_el < s->fp_excp_el) { + s->fp_access_checked = true; + return sme_access_check(s); + } + return fp_access_check_only(s); +} + +/* Common subroutine for CheckSMEAnd*Enabled. */ +bool sme_enabled_check_with_svcr(DisasContext *s, unsigned req) +{ + if (!sme_enabled_check(s)) { + return false; + } + if (FIELD_EX64(req, SVCR, SM) && !s->pstate_sm) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_NotStreaming, false)); + return false; + } + if (FIELD_EX64(req, SVCR, ZA) && !s->pstate_za) { + gen_exception_insn(s, s->pc_curr, EXCP_UDEF, + syn_smetrap(SME_ET_InactiveZA, false)); + return false; + } + return true; +} + /* * This utility function is for doing register extension with an * optional shift. You will likely want to pass a temporary for the -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> The pseudocode for CheckSVEEnabled gains a check for Streaming SVE mode, and for SME present but SVE absent. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-17-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.c +++ b/target/arm/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool fp_access_check(DisasContext *s) return true; } -/* Check that SVE access is enabled. If it is, return true. +/* + * Check that SVE access is enabled. If it is, return true. * If not, emit code to generate an appropriate exception and return false. + * This function corresponds to CheckSVEEnabled(). */ bool sve_access_check(DisasContext *s) { - if (s->sve_excp_el) { - assert(!s->sve_access_checked); - s->sve_access_checked = true; - + if (s->pstate_sm || !dc_isar_feature(aa64_sve, s)) { + assert(dc_isar_feature(aa64_sme, s)); + if (!sme_sm_enabled_check(s)) { + goto fail_exit; + } + } else if (s->sve_excp_el) { gen_exception_insn_el(s, s->pc_curr, EXCP_UDEF, syn_sve_access_trap(), s->sve_excp_el); - return false; + goto fail_exit; } s->sve_access_checked = true; return fp_access_check(s); + + fail_exit: + /* Assert that we only raise one exception per instruction. */ + assert(!s->sve_access_checked); + s->sve_access_checked = true; + return false; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These SME instructions are nominally within the SVE decode space, so we add them to sve.decode and translate-sve.c. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-18-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 12 ++++++++++++ target/arm/sve.decode | 5 ++++- target/arm/translate-sve.c | 38 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 54 insertions(+), 1 deletion(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int vec_full_reg_size(DisasContext *s) return s->vl; } +/* Return the byte size of the vector register, SVL / 8. */ +static inline int streaming_vec_reg_size(DisasContext *s) +{ + return s->svl; +} + /* * Return the offset info CPUARMState of the predicate vector register Pn. * Note for this purpose, FFR is P16. @@ -XXX,XX +XXX,XX @@ static inline int pred_full_reg_size(DisasContext *s) return s->vl >> 3; } +/* Return the byte size of the predicate register, SVL / 64. */ +static inline int streaming_pred_reg_size(DisasContext *s) +{ + return s->svl >> 3; +} + /* * Round up the size of a register to a size allowed by * the tcg vector infrastructure. Any operation which uses this diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ INDEX_ri 00000100 esz:2 1 imm:s5 010001 rn:5 rd:5 # SVE index generation (register start, register increment) INDEX_rr 00000100 .. 1 ..... 010011 ..... ..... @rd_rn_rm -### SVE Stack Allocation Group +### SVE / Streaming SVE Stack Allocation Group # SVE stack frame adjustment ADDVL 00000100 001 ..... 01010 ...... ..... @rd_rn_i6 +ADDSVL 00000100 001 ..... 01011 ...... ..... @rd_rn_i6 ADDPL 00000100 011 ..... 01010 ...... ..... @rd_rn_i6 +ADDSPL 00000100 011 ..... 01011 ...... ..... @rd_rn_i6 # SVE stack frame size RDVL 00000100 101 11111 01010 imm:s6 rd:5 +RDSVL 00000100 101 11111 01011 imm:s6 rd:5 ### SVE Bitwise Shift - Unpredicated Group diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_ADDVL(DisasContext *s, arg_ADDVL *a) return true; } +static bool trans_ADDSVL(DisasContext *s, arg_ADDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ADDPL(DisasContext *s, arg_ADDPL *a) return true; } +static bool trans_ADDSPL(DisasContext *s, arg_ADDSPL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 rd = cpu_reg_sp(s, a->rd); + TCGv_i64 rn = cpu_reg_sp(s, a->rn); + tcg_gen_addi_i64(rd, rn, a->imm * streaming_pred_reg_size(s)); + } + return true; +} + static bool trans_RDVL(DisasContext *s, arg_RDVL *a) { if (!dc_isar_feature(aa64_sve, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_RDVL(DisasContext *s, arg_RDVL *a) return true; } +static bool trans_RDSVL(DisasContext *s, arg_RDSVL *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_enabled_check(s)) { + TCGv_i64 reg = cpu_reg(s, a->rd); + tcg_gen_movi_i64(reg, a->imm * streaming_vec_reg_size(s)); + } + return true; +} + /* *** SVE Compute Vector Address Group */ -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-19-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 4 ++++ target/arm/sme_helper.c | 25 +++++++++++++++++++++++++ target/arm/translate-sme.c | 13 +++++++++++++ 4 files changed, 44 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) + +DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ # # This file is processed by scripts/decodetree.py # + +### SME Misc + +ZERO 11000000 00 001 00000000000 imm:8 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void helper_set_pstate_za(CPUARMState *env, uint32_t i) memset(env->zarray, 0, sizeof(env->zarray)); } } + +void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) +{ + uint32_t i; + + /* + * Special case clearing the entire ZA space. + * This falls into the CONSTRAINED UNPREDICTABLE zeroing of any + * parts of the ZA storage outside of SVL. + */ + if (imm == 0xff) { + memset(env->zarray, 0, sizeof(env->zarray)); + return; + } + + /* + * Recall that ZAnH.D[m] is spread across ZA[n+8*m], + * so each row is discontiguous within ZA[]. + */ + for (i = 0; i < svl; i++) { + if (imm & (1 << (i % 8))) { + memset(&env->zarray[i], 0, svl); + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ */ #include "decode-sme.c.inc" + + +static bool trans_ZERO(DisasContext *s, arg_ZERO *a) +{ + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (sme_za_enabled_check(s)) { + gen_helper_sme_zero(cpu_env, tcg_constant_i32(a->imm), + tcg_constant_i32(streaming_vec_reg_size(s))); + } + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for implementing moves to/from horizontal tile slices, but we need new ones for moves to/from vertical tile slices. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-20-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 12 +++ target/arm/helper-sve.h | 2 + target/arm/translate-a64.h | 8 ++ target/arm/translate.h | 5 ++ target/arm/sme.decode | 15 ++++ target/arm/sme_helper.c | 151 ++++++++++++++++++++++++++++++++++++- target/arm/sve_helper.c | 12 +++ target/arm/translate-sme.c | 127 +++++++++++++++++++++++++++++++ 8 files changed, 331 insertions(+), 1 deletion(-) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(set_pstate_sm, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_2(set_pstate_za, TCG_CALL_NO_RWG, void, env, i32) DEF_HELPER_FLAGS_3(sme_zero, TCG_CALL_NO_RWG, void, env, i32, i32) + +/* Move to/from vertical array slices, i.e. columns, so 'c'. */ +DEF_HELPER_FLAGS_4(sme_mova_cz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sve_sel_zpzz_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve_sel_zpzz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sve_sel_zpzz_q, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sve2_addp_zpzz_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ static inline int pred_gvec_reg_size(DisasContext *s) return size_for_gvec(pred_full_reg_size(s)); } +/* Return a newly allocated pointer to the predicate register. */ +static inline TCGv_ptr pred_full_reg_ptr(DisasContext *s, int regno) +{ + TCGv_ptr ret = tcg_temp_new_ptr(); + tcg_gen_addi_ptr(ret, cpu_env, pred_full_reg_offset(s, regno)); + return ret; +} + bool disas_sve(DisasContext *, uint32_t); bool disas_sme(DisasContext *, uint32_t); diff --git a/target/arm/translate.h b/target/arm/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.h +++ b/target/arm/translate.h @@ -XXX,XX +XXX,XX @@ static inline int plus_2(DisasContext *s, int x) return x + 2; } +static inline int plus_12(DisasContext *s, int x) +{ + return x + 12; +} + static inline int times_2(DisasContext *s, int x) { return x * 2; diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ### SME Misc ZERO 11000000 00 001 00000000000 imm:8 + +### SME Move into/from Array + +%mova_rs 13:2 !function=plus_12 +&mova esz rs pg zr za_imm v:bool to_vec:bool + +MOVA 11000000 esz:2 00000 0 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs +MOVA 11000000 11 00000 1 v:1 .. pg:3 zr:5 0 za_imm:4 \ + &mova to_vec=0 rs=%mova_rs esz=4 + +MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs +MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ + &mova to_vec=1 rs=%mova_rs esz=4 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" -#include "internals.h" +#include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "qemu/int128.h" +#include "vec_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void helper_sme_zero(CPUARMState *env, uint32_t imm, uint32_t svl) } } } + + +/* + * When considering the ZA storage as an array of elements of + * type T, the index within that array of the Nth element of + * a vertical slice of a tile can be calculated like this, + * regardless of the size of type T. This is because the tiles + * are interleaved, so if type T is size N bytes then row 1 of + * the tile is N rows away from row 0. The division by N to + * convert a byte offset into an array index and the multiplication + * by N to convert from vslice-index-within-the-tile to + * the index within the ZA storage cancel out. + */ +#define tile_vslice_index(i) ((i) * sizeof(ARMVectorReg)) + +/* + * When doing byte arithmetic on the ZA storage, the element + * byteoff bytes away in a tile vertical slice is always this + * many bytes away in the ZA storage, regardless of the + * size of the tile element, assuming that byteoff is a multiple + * of the element size. Again this is because of the interleaving + * of the tiles. For instance if we have 1 byte per element then + * each row of the ZA storage has one byte of the vslice data, + * and (counting from 0) byte 8 goes in row 8 of the storage + * at offset (8 * row-size-in-bytes). + * If we have 8 bytes per element then each row of the ZA storage + * has 8 bytes of the data, but there are 8 interleaved tiles and + * so byte 8 of the data goes into row 1 of the tile, + * which is again row 8 of the storage, so the offset is still + * (8 * row-size-in-bytes). Similarly for other element sizes. + */ +#define tile_vslice_offset(byteoff) ((byteoff) * sizeof(ARMVectorReg)) + + +/* + * Move Zreg vector to ZArray column. + */ +#define DO_MOVA_C(NAME, TYPE, H) \ +void HELPER(NAME)(void *za, void *vn, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(za + tile_vslice_offset(i)) = *(TYPE *)(vn + H(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_C(sme_mova_cz_b, uint8_t, H1) +DO_MOVA_C(sme_mova_cz_h, uint16_t, H1_2) +DO_MOVA_C(sme_mova_cz_s, uint32_t, H1_4) + +void HELPER(sme_mova_cz_d)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *n = vn; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +void HELPER(sme_mova_cz_q)(void *za, void *vn, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *n = vn; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++) { + if (pg[H2(i)] & 1) { + a[tile_vslice_index(i)] = n[i]; + } + } +} + +#undef DO_MOVA_C + +/* + * Move ZArray column to Zreg vector. + */ +#define DO_MOVA_Z(NAME, TYPE, H) \ +void HELPER(NAME)(void *vd, void *za, void *vg, uint32_t desc) \ +{ \ + int i, oprsz = simd_oprsz(desc); \ + for (i = 0; i < oprsz; ) { \ + uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ + do { \ + if (pg & 1) { \ + *(TYPE *)(vd + H(i)) = *(TYPE *)(za + tile_vslice_offset(i)); \ + } \ + i += sizeof(TYPE); \ + pg >>= sizeof(TYPE); \ + } while (i & 15); \ + } \ +} + +DO_MOVA_Z(sme_mova_zc_b, uint8_t, H1) +DO_MOVA_Z(sme_mova_zc_h, uint16_t, H1_2) +DO_MOVA_Z(sme_mova_zc_s, uint32_t, H1_4) + +void HELPER(sme_mova_zc_d)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 8; + uint8_t *pg = vg; + uint64_t *d = vd; + uint64_t *a = za; + + for (i = 0; i < oprsz; i++) { + if (pg[H1(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) +{ + int i, oprsz = simd_oprsz(desc) / 16; + uint16_t *pg = vg; + Int128 *d = vd; + Int128 *a = za; + + /* + * Int128 is used here simply to copy 16 bytes, and to simplify + * the address arithmetic. + */ + for (i = 0; i < oprsz; i++, za += sizeof(ARMVectorReg)) { + if (pg[H2(i)] & 1) { + d[i] = a[tile_vslice_index(i)]; + } + } +} + +#undef DO_MOVA_Z diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve_helper.c +++ b/target/arm/sve_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sve_sel_zpzz_d)(void *vd, void *vn, void *vm, } } +void HELPER(sve_sel_zpzz_q)(void *vd, void *vn, void *vm, + void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 16; + Int128 *d = vd, *n = vn, *m = vm; + uint16_t *pg = vg; + + for (i = 0; i < opr_sz; i += 1) { + d[i] = (pg[H2(i)] & 1 ? n : m)[i]; + } +} + /* Two operand comparison controlled by a predicate. * ??? It is very tempting to want to be able to expand this inline * with x86 instructions, e.g. diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ #include "decode-sme.c.inc" +/* + * Resolve tile.size[index] to a host pointer, where tile and index + * are always decoded together, dependent on the element size. + */ +static TCGv_ptr get_tile_rowcol(DisasContext *s, int esz, int rs, + int tile_index, bool vertical) +{ + int tile = tile_index >> (4 - esz); + int index = esz == MO_128 ? 0 : extract32(tile_index, 0, 4 - esz); + int pos, len, offset; + TCGv_i32 tmp; + TCGv_ptr addr; + + /* Compute the final index, which is Rs+imm. */ + tmp = tcg_temp_new_i32(); + tcg_gen_trunc_tl_i32(tmp, cpu_reg(s, rs)); + tcg_gen_addi_i32(tmp, tmp, index); + + /* Prepare a power-of-two modulo via extraction of @len bits. */ + len = ctz32(streaming_vec_reg_size(s)) - esz; + + if (vertical) { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * size + * = (index % (svl >> esz)) << esz + * Perform the power-of-two modulo via extraction of the low @len bits. + * Perform the multiply by shifting left by @pos bits. + * Perform these operations simultaneously via deposit into zero. + */ + pos = esz; + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* + * For big-endian, adjust the indexed column byte offset within + * the uint64_t host words that make up env->zarray[]. + */ + if (HOST_BIG_ENDIAN && esz < MO_64) { + tcg_gen_xori_i32(tmp, tmp, 8 - (1 << esz)); + } + } else { + /* + * Compute the byte offset of the index within the tile: + * (index % (svl / size)) * (size * sizeof(row)) + * = (index % (svl >> esz)) << (esz + log2(sizeof(row))) + */ + pos = esz + ctz32(sizeof(ARMVectorReg)); + tcg_gen_deposit_z_i32(tmp, tmp, pos, len); + + /* Row slices are always aligned and need no endian adjustment. */ + } + + /* The tile byte offset within env->zarray is the row. */ + offset = tile * sizeof(ARMVectorReg); + + /* Include the byte offset of zarray to make this relative to env. */ + offset += offsetof(CPUARMState, zarray); + tcg_gen_addi_i32(tmp, tmp, offset); + + /* Add the byte offset to env to produce the final pointer. */ + addr = tcg_temp_new_ptr(); + tcg_gen_ext_i32_ptr(addr, tmp); + tcg_temp_free_i32(tmp); + tcg_gen_add_ptr(addr, addr, cpu_env); + + return addr; +} + static bool trans_ZERO(DisasContext *s, arg_ZERO *a) { if (!dc_isar_feature(aa64_sme, s)) { @@ -XXX,XX +XXX,XX @@ static bool trans_ZERO(DisasContext *s, arg_ZERO *a) } return true; } + +static bool trans_MOVA(DisasContext *s, arg_MOVA *a) +{ + static gen_helper_gvec_4 * const h_fns[5] = { + gen_helper_sve_sel_zpzz_b, gen_helper_sve_sel_zpzz_h, + gen_helper_sve_sel_zpzz_s, gen_helper_sve_sel_zpzz_d, + gen_helper_sve_sel_zpzz_q + }; + static gen_helper_gvec_3 * const cz_fns[5] = { + gen_helper_sme_mova_cz_b, gen_helper_sme_mova_cz_h, + gen_helper_sme_mova_cz_s, gen_helper_sme_mova_cz_d, + gen_helper_sme_mova_cz_q, + }; + static gen_helper_gvec_3 * const zc_fns[5] = { + gen_helper_sme_mova_zc_b, gen_helper_sme_mova_zc_h, + gen_helper_sme_mova_zc_s, gen_helper_sme_mova_zc_d, + gen_helper_sme_mova_zc_q, + }; + + TCGv_ptr t_za, t_zr, t_pg; + TCGv_i32 t_desc; + int svl; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_zr = vec_full_reg_ptr(s, a->zr); + t_pg = pred_full_reg_ptr(s, a->pg); + + svl = streaming_vec_reg_size(s); + t_desc = tcg_constant_i32(simd_desc(svl, svl, 0)); + + if (a->v) { + /* Vertical slice -- use sme mova helpers. */ + if (a->to_vec) { + zc_fns[a->esz](t_zr, t_za, t_pg, t_desc); + } else { + cz_fns[a->esz](t_za, t_zr, t_pg, t_desc); + } + } else { + /* Horizontal slice -- reuse sve sel helpers. */ + if (a->to_vec) { + h_fns[a->esz](t_zr, t_za, t_zr, t_pg, t_desc); + } else { + h_fns[a->esz](t_za, t_zr, t_za, t_pg, t_desc); + } + } + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_zr); + tcg_temp_free_ptr(t_pg); + + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We cannot reuse the SVE functions for LD[1-4] and ST[1-4], because those functions accept only a Zreg register number. For SME, we want to pass a pointer into ZA storage. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-21-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 82 +++++ target/arm/sme.decode | 9 + target/arm/sme_helper.c | 595 +++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 70 +++++ 4 files changed, 756 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sme_mova_cz_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_cz_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sme_mova_zc_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(sme_ld1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_ld1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_ld1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1b_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1b_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1h_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1h_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1s_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1s_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1d_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1d_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_st1q_be_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) +DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ MOVA 11000000 esz:2 00001 0 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs MOVA 11000000 11 00001 1 v:1 .. pg:3 0 za_imm:4 zr:5 \ &mova to_vec=1 rs=%mova_rs esz=4 + +### SME Memory + +&ldst esz rs pg rn rm za_imm v:bool st:bool + +LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst rs=%mova_rs +LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ + &ldst esz=4 rs=%mova_rs diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "cpu.h" +#include "internals.h" #include "tcg/tcg-gvec-desc.h" #include "exec/helper-proto.h" +#include "exec/cpu_ldst.h" +#include "exec/exec-all.h" #include "qemu/int128.h" #include "vec_internal.h" +#include "sve_ldst_internal.h" /* ResetSVEState */ void arm_reset_sve_state(CPUARMState *env) @@ -XXX,XX +XXX,XX @@ void HELPER(sme_mova_zc_q)(void *vd, void *za, void *vg, uint32_t desc) } #undef DO_MOVA_Z + +/* + * Clear elements in a tile slice comprising len bytes. + */ + +typedef void ClearFn(void *ptr, size_t off, size_t len); + +static void clear_horizontal(void *ptr, size_t off, size_t len) +{ + memset(ptr + off, 0, len); +} + +static void clear_vertical_b(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; ++i) { + *(uint8_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_h(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 2) { + *(uint16_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_s(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 4) { + *(uint32_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_d(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 8) { + *(uint64_t *)(vptr + tile_vslice_offset(i + off)) = 0; + } +} + +static void clear_vertical_q(void *vptr, size_t off, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memset(vptr + tile_vslice_offset(i + off), 0, 16); + } +} + +/* + * Copy elements from an array into a tile slice comprising len bytes. + */ + +typedef void CopyFn(void *dst, const void *src, size_t len); + +static void copy_horizontal(void *dst, const void *src, size_t len) +{ + memcpy(dst, src, len); +} + +static void copy_vertical_b(void *vdst, const void *vsrc, size_t len) +{ + const uint8_t *src = vsrc; + uint8_t *dst = vdst; + size_t i; + + for (i = 0; i < len; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_h(void *vdst, const void *vsrc, size_t len) +{ + const uint16_t *src = vsrc; + uint16_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 2; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_s(void *vdst, const void *vsrc, size_t len) +{ + const uint32_t *src = vsrc; + uint32_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 4; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_d(void *vdst, const void *vsrc, size_t len) +{ + const uint64_t *src = vsrc; + uint64_t *dst = vdst; + size_t i; + + for (i = 0; i < len / 8; ++i) { + dst[tile_vslice_index(i)] = src[i]; + } +} + +static void copy_vertical_q(void *vdst, const void *vsrc, size_t len) +{ + for (size_t i = 0; i < len; i += 16) { + memcpy(vdst + tile_vslice_offset(i), vsrc + i, 16); + } +} + +/* + * Host and TLB primitives for vertical tile slice addressing. + */ + +#define DO_LD(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = HOST(host); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = TLB(env, useronly_clean_ptr(addr), ra); \ + *(TYPE *)(za + tile_vslice_offset(off)) = val; \ +} + +#define DO_ST(NAME, TYPE, HOST, TLB) \ +static inline void sme_##NAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + HOST(host, val); \ +} \ +static inline void sme_##NAME##_v_tlb(CPUARMState *env, void *za, \ + intptr_t off, target_ulong addr, uintptr_t ra) \ +{ \ + TYPE val = *(TYPE *)(za + tile_vslice_offset(off)); \ + TLB(env, useronly_clean_ptr(addr), val, ra); \ +} + +/* + * The ARMVectorReg elements are stored in host-endian 64-bit units. + * For 128-bit quantities, the sequence defined by the Elem[] pseudocode + * corresponds to storing the two 64-bit pieces in little-endian order. + */ +#define DO_LDQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t val0 = HOST(host), val1 = HOST(host + 8); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t val0 = TLB(env, useronly_clean_ptr(addr), ra); \ + uint64_t val1 = TLB(env, useronly_clean_ptr(addr + 8), ra); \ + uint64_t *ptr = za + off; \ + ptr[0] = BE ? val1 : val0, ptr[1] = BE ? val0 : val1; \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +#define DO_STQ(HNAME, VNAME, BE, HOST, TLB) \ +static inline void HNAME##_host(void *za, intptr_t off, void *host) \ +{ \ + uint64_t *ptr = za + off; \ + HOST(host, ptr[BE]); \ + HOST(host + 1, ptr[!BE]); \ +} \ +static inline void VNAME##_v_host(void *za, intptr_t off, void *host) \ +{ \ + HNAME##_host(za, tile_vslice_offset(off), host); \ +} \ +static inline void HNAME##_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + uint64_t *ptr = za + off; \ + TLB(env, useronly_clean_ptr(addr), ptr[BE], ra); \ + TLB(env, useronly_clean_ptr(addr + 8), ptr[!BE], ra); \ +} \ +static inline void VNAME##_v_tlb(CPUARMState *env, void *za, intptr_t off, \ + target_ulong addr, uintptr_t ra) \ +{ \ + HNAME##_tlb(env, za, tile_vslice_offset(off), addr, ra); \ +} + +DO_LD(ld1b, uint8_t, ldub_p, cpu_ldub_data_ra) +DO_LD(ld1h_be, uint16_t, lduw_be_p, cpu_lduw_be_data_ra) +DO_LD(ld1h_le, uint16_t, lduw_le_p, cpu_lduw_le_data_ra) +DO_LD(ld1s_be, uint32_t, ldl_be_p, cpu_ldl_be_data_ra) +DO_LD(ld1s_le, uint32_t, ldl_le_p, cpu_ldl_le_data_ra) +DO_LD(ld1d_be, uint64_t, ldq_be_p, cpu_ldq_be_data_ra) +DO_LD(ld1d_le, uint64_t, ldq_le_p, cpu_ldq_le_data_ra) + +DO_LDQ(sve_ld1qq_be, sme_ld1q_be, 1, ldq_be_p, cpu_ldq_be_data_ra) +DO_LDQ(sve_ld1qq_le, sme_ld1q_le, 0, ldq_le_p, cpu_ldq_le_data_ra) + +DO_ST(st1b, uint8_t, stb_p, cpu_stb_data_ra) +DO_ST(st1h_be, uint16_t, stw_be_p, cpu_stw_be_data_ra) +DO_ST(st1h_le, uint16_t, stw_le_p, cpu_stw_le_data_ra) +DO_ST(st1s_be, uint32_t, stl_be_p, cpu_stl_be_data_ra) +DO_ST(st1s_le, uint32_t, stl_le_p, cpu_stl_le_data_ra) +DO_ST(st1d_be, uint64_t, stq_be_p, cpu_stq_be_data_ra) +DO_ST(st1d_le, uint64_t, stq_le_p, cpu_stq_le_data_ra) + +DO_STQ(sve_st1qq_be, sme_st1q_be, 1, stq_be_p, cpu_stq_be_data_ra) +DO_STQ(sve_st1qq_le, sme_st1q_le, 0, stq_le_p, cpu_stq_le_data_ra) + +#undef DO_LD +#undef DO_ST +#undef DO_LDQ +#undef DO_STQ + +/* + * Common helper for all contiguous predicated loads. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_ld1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no load occurs. */ + clr_fn(za, 0, reg_max); + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_LOAD, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_READ, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. Perform the load + * into scratch memory to preserve register state until the end. + */ + ARMVectorReg scratch = { }; + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, &scratch, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + + cpy_fn(za, &scratch, reg_max); + return; +#endif + } + + /* The entire operation is in RAM, on valid pages. */ + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + if (!vertical) { + memset(za, 0, reg_max); + } else if (reg_off) { + clr_fn(za, 0, reg_off); + } + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } else if (vertical) { + clr_fn(za, reg_off, esize); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_ld1_mte(CPUARMState *env, void *za, uint64_t *vg, + target_ulong addr, uint32_t desc, uintptr_t ra, + const int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn, + ClearFn *clr_fn, + CopyFn *cpy_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_ld1(env, za, vg, addr, desc, ra, esz, mtedesc, vertical, + host_fn, tlb_fn, clr_fn, cpy_fn); +} + +#define DO_LD(L, END, ESZ) \ +void HELPER(sme_ld1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} \ +void HELPER(sme_ld1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_ld1##L##L##END##_host, sve_ld1##L##L##END##_tlb, \ + clear_horizontal, copy_horizontal); \ +} \ +void HELPER(sme_ld1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_ld1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_ld1##L##END##_v_host, sme_ld1##L##END##_v_tlb, \ + clear_vertical_##L, copy_vertical_##L); \ +} + +DO_LD(b, , MO_8) +DO_LD(h, _be, MO_16) +DO_LD(h, _le, MO_16) +DO_LD(s, _be, MO_32) +DO_LD(s, _le, MO_32) +DO_LD(d, _be, MO_64) +DO_LD(d, _le, MO_64) +DO_LD(q, _be, MO_128) +DO_LD(q, _le, MO_128) + +#undef DO_LD + +/* + * Common helper for all contiguous predicated stores. + */ + +static inline QEMU_ALWAYS_INLINE +void sme_st1(CPUARMState *env, void *za, uint64_t *vg, + const target_ulong addr, uint32_t desc, const uintptr_t ra, + const int esz, uint32_t mtedesc, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + const intptr_t reg_max = simd_oprsz(desc); + const intptr_t esize = 1 << esz; + intptr_t reg_off, reg_last; + SVEContLdSt info; + void *host; + int flags; + + /* Find the active elements. */ + if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, esize)) { + /* The entire predicate was false; no store occurs. */ + return; + } + + /* Probe the page(s). Exit with exception for any invalid page. */ + sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_STORE, ra); + + /* Handle watchpoints for all active elements. */ + sve_cont_ldst_watchpoints(&info, env, vg, addr, esize, esize, + BP_MEM_WRITE, ra); + + /* + * Handle mte checks for all active elements. + * Since TBI must be set for MTE, !mtedesc => !mte_active. + */ + if (mtedesc) { + sve_cont_ldst_mte_check(&info, env, vg, addr, esize, esize, + mtedesc, ra); + } + + flags = info.page[0].flags | info.page[1].flags; + if (unlikely(flags != 0)) { +#ifdef CONFIG_USER_ONLY + g_assert_not_reached(); +#else + /* + * At least one page includes MMIO. + * Any bus operation can fail with cpu_transaction_failed, + * which for ARM will raise SyncExternal. We cannot avoid + * this fault and will leave with the store incomplete. + */ + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[1]; + if (reg_last < 0) { + reg_last = info.reg_off_split; + if (reg_last < 0) { + reg_last = info.reg_off_last[0]; + } + } + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + reg_off += esize; + } while (reg_off & 63); + } while (reg_off <= reg_last); + return; +#endif + } + + reg_off = info.reg_off_first[0]; + reg_last = info.reg_off_last[0]; + host = info.page[0].host; + + while (reg_off <= reg_last) { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off <= reg_last && (reg_off & 63)); + } + + /* + * Use the slow path to manage the cross-page misalignment. + * But we know this is RAM and cannot trap. + */ + reg_off = info.reg_off_split; + if (unlikely(reg_off >= 0)) { + tlb_fn(env, za, reg_off, addr + reg_off, ra); + } + + reg_off = info.reg_off_first[1]; + if (unlikely(reg_off >= 0)) { + reg_last = info.reg_off_last[1]; + host = info.page[1].host; + + do { + uint64_t pg = vg[reg_off >> 6]; + do { + if ((pg >> (reg_off & 63)) & 1) { + host_fn(za, reg_off, host + reg_off); + } + reg_off += 1 << esz; + } while (reg_off & 63); + } while (reg_off <= reg_last); + } +} + +static inline QEMU_ALWAYS_INLINE +void sme_st1_mte(CPUARMState *env, void *za, uint64_t *vg, target_ulong addr, + uint32_t desc, uintptr_t ra, int esz, bool vertical, + sve_ldst1_host_fn *host_fn, + sve_ldst1_tlb_fn *tlb_fn) +{ + uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + int bit55 = extract64(addr, 55, 1); + + /* Remove mtedesc from the normal sve descriptor. */ + desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); + + /* Perform gross MTE suppression early. */ + if (!tbi_check(desc, bit55) || + tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { + mtedesc = 0; + } + + sme_st1(env, za, vg, addr, desc, ra, esz, mtedesc, + vertical, host_fn, tlb_fn); +} + +#define DO_ST(L, END, ESZ) \ +void HELPER(sme_st1##L##END##_h)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1(env, za, vg, addr, desc, GETPC(), ESZ, 0, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} \ +void HELPER(sme_st1##L##END##_h_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, false, \ + sve_st1##L##L##END##_host, sve_st1##L##L##END##_tlb); \ +} \ +void HELPER(sme_st1##L##END##_v_mte)(CPUARMState *env, void *za, void *vg, \ + target_ulong addr, uint32_t desc) \ +{ \ + sme_st1_mte(env, za, vg, addr, desc, GETPC(), ESZ, true, \ + sme_st1##L##END##_v_host, sme_st1##L##END##_v_tlb); \ +} + +DO_ST(b, , MO_8) +DO_ST(h, _be, MO_16) +DO_ST(h, _le, MO_16) +DO_ST(s, _be, MO_32) +DO_ST(s, _le, MO_32) +DO_ST(d, _be, MO_64) +DO_ST(d, _le, MO_64) +DO_ST(q, _be, MO_128) +DO_ST(q, _le, MO_128) + +#undef DO_ST diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_MOVA(DisasContext *s, arg_MOVA *a) return true; } + +static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) +{ + typedef void GenLdSt1(TCGv_env, TCGv_ptr, TCGv_ptr, TCGv, TCGv_i32); + + /* + * Indexed by [esz][be][v][mte][st], which is (except for load/store) + * also the order in which the elements appear in the function names, + * and so how we must concatenate the pieces. + */ + +#define FN_LS(F) { gen_helper_sme_ld1##F, gen_helper_sme_st1##F } +#define FN_MTE(F) { FN_LS(F), FN_LS(F##_mte) } +#define FN_HV(F) { FN_MTE(F##_h), FN_MTE(F##_v) } +#define FN_END(L, B) { FN_HV(L), FN_HV(B) } + + static GenLdSt1 * const fns[5][2][2][2][2] = { + FN_END(b, b), + FN_END(h_le, h_be), + FN_END(s_le, s_be), + FN_END(d_le, d_be), + FN_END(q_le, q_be), + }; + +#undef FN_LS +#undef FN_MTE +#undef FN_HV +#undef FN_END + + TCGv_ptr t_za, t_pg; + TCGv_i64 addr; + int svl, desc = 0; + bool be = s->be_data == MO_BE; + bool mte = s->mte_active[0]; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sme_smza_enabled_check(s)) { + return true; + } + + t_za = get_tile_rowcol(s, a->esz, a->rs, a->za_imm, a->v); + t_pg = pred_full_reg_ptr(s, a->pg); + addr = tcg_temp_new_i64(); + + tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), a->esz); + tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, a->rn)); + + if (mte) { + desc = FIELD_DP32(desc, MTEDESC, MIDX, get_mem_index(s)); + desc = FIELD_DP32(desc, MTEDESC, TBI, s->tbid); + desc = FIELD_DP32(desc, MTEDESC, TCMA, s->tcma); + desc = FIELD_DP32(desc, MTEDESC, WRITE, a->st); + desc = FIELD_DP32(desc, MTEDESC, SIZEM1, (1 << a->esz) - 1); + desc <<= SVE_MTEDESC_SHIFT; + } else { + addr = clean_data_tbi(s, addr); + } + svl = streaming_vec_reg_size(s); + desc = simd_desc(svl, svl, desc); + + fns[a->esz][be][a->v][mte][a->st](cpu_env, t_za, t_pg, addr, + tcg_constant_i32(desc)); + + tcg_temp_free_ptr(t_za); + tcg_temp_free_ptr(t_pg); + tcg_temp_free_i64(addr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add a TCGv_ptr base argument, which will be cpu_env for SVE. We will reuse this for SME save and restore array insns. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-22-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate-a64.h | 3 +++ target/arm/translate-sve.c | 48 ++++++++++++++++++++++++++++---------- 2 files changed, 39 insertions(+), 12 deletions(-) diff --git a/target/arm/translate-a64.h b/target/arm/translate-a64.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-a64.h +++ b/target/arm/translate-a64.h @@ -XXX,XX +XXX,XX @@ void gen_gvec_xar(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, uint32_t rm_ofs, int64_t shift, uint32_t opr_sz, uint32_t max_sz); +void gen_sve_ldr(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); +void gen_sve_str(DisasContext *s, TCGv_ptr, int vofs, int len, int rn, int imm); + #endif /* TARGET_ARM_TRANSLATE_A64_H */ diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(UCVTF_dd, aa64_sve, gen_gvec_fpst_arg_zpz, * The load should begin at the address Rn + IMM. */ -static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_ldr(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { tcg_gen_qemu_ld_i64(t0, clean_addr, midx, MO_LEUQ); - tcg_gen_st_i64(t0, cpu_env, vofs + i); + tcg_gen_st_i64(t0, base, vofs + i); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } tcg_temp_free_i64(t0); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_addi_i64(clean_addr, clean_addr, 8); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_addi_ptr(i, i, 8); tcg_gen_st_i64(t0, tp, vofs); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* @@ -XXX,XX +XXX,XX @@ static void do_ldr(DisasContext *s, uint32_t vofs, int len, int rn, int imm) default: g_assert_not_reached(); } - tcg_gen_st_i64(t0, cpu_env, vofs + len_align); + tcg_gen_st_i64(t0, base, vofs + len_align); tcg_temp_free_i64(t0); } } /* Similarly for stores. */ -static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) +void gen_sve_str(DisasContext *s, TCGv_ptr base, int vofs, + int len, int rn, int imm) { int len_align = QEMU_ALIGN_DOWN(len, 8); int len_remain = len % 8; @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) t0 = tcg_temp_new_i64(); for (i = 0; i < len_align; i += 8) { - tcg_gen_ld_i64(t0, cpu_env, vofs + i); + tcg_gen_ld_i64(t0, base, vofs + i); tcg_gen_qemu_st_i64(t0, clean_addr, midx, MO_LEUQ); tcg_gen_addi_i64(clean_addr, clean_addr, 8); } @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) clean_addr = new_tmp_a64_local(s); tcg_gen_mov_i64(clean_addr, t0); + if (base != cpu_env) { + TCGv_ptr b = tcg_temp_local_new_ptr(); + tcg_gen_mov_ptr(b, base); + base = b; + } + gen_set_label(loop); t0 = tcg_temp_new_i64(); tp = tcg_temp_new_ptr(); - tcg_gen_add_ptr(tp, cpu_env, i); + tcg_gen_add_ptr(tp, base, i); tcg_gen_ld_i64(t0, tp, vofs); tcg_gen_addi_ptr(i, i, 8); tcg_temp_free_ptr(tp); @@ -XXX,XX +XXX,XX @@ static void do_str(DisasContext *s, uint32_t vofs, int len, int rn, int imm) tcg_gen_brcondi_ptr(TCG_COND_LTU, i, len_align, loop); tcg_temp_free_ptr(i); + + if (base != cpu_env) { + tcg_temp_free_ptr(base); + assert(len_remain == 0); + } } /* Predicate register stores can be any multiple of 2. */ if (len_remain) { t0 = tcg_temp_new_i64(); - tcg_gen_ld_i64(t0, cpu_env, vofs + len_align); + tcg_gen_ld_i64(t0, base, vofs + len_align); switch (len_remain) { case 2: @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_LDR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_ldr(s, off, size, a->rn, a->imm * size); + gen_sve_ldr(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_zri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = vec_full_reg_size(s); int off = vec_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } @@ -XXX,XX +XXX,XX @@ static bool trans_STR_pri(DisasContext *s, arg_rri *a) if (sve_access_check(s)) { int size = pred_full_reg_size(s); int off = pred_full_reg_offset(s, a->rd); - do_str(s, off, size, a->rn, a->imm * size); + gen_sve_str(s, cpu_env, off, size, a->rn, a->imm * size); } return true; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can reuse the SVE functions for LDR and STR, passing in the base of the ZA vector and a zero offset. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-23-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sme.decode | 7 +++++++ target/arm/translate-sme.c | 24 ++++++++++++++++++++++++ 2 files changed, 31 insertions(+) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 0 esz:2 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst rs=%mova_rs LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ &ldst esz=4 rs=%mova_rs + +&ldstr rv rn imm +@ldstr ....... ... . ...... .. ... rn:5 . imm:4 \ + &ldstr rv=%mova_rs + +LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr +STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool trans_LDST1(DisasContext *s, arg_LDST1 *a) tcg_temp_free_i64(addr); return true; } + +typedef void GenLdStR(DisasContext *, TCGv_ptr, int, int, int, int); + +static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) +{ + int svl = streaming_vec_reg_size(s); + int imm = a->imm; + TCGv_ptr base; + + if (!sme_za_enabled_check(s)) { + return true; + } + + /* ZA[n] equates to ZA0H.B[n]. */ + base = get_tile_rowcol(s, MO_8, a->rv, imm, false); + + fn(s, base, 0, svl, a->rn, imm * svl); + + tcg_temp_free_ptr(base); + return true; +} + +TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) +TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-24-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 11 +++++ target/arm/sme_helper.c | 90 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 31 +++++++++++++ 4 files changed, 137 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_st1q_be_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i DEF_HELPER_FLAGS_5(sme_st1q_le_h_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_be_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) DEF_HELPER_FLAGS_5(sme_st1q_le_v_mte, TCG_CALL_NO_WG, void, env, ptr, ptr, tl, i32) + +DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ LDST1 1110000 111 st:1 rm:5 v:1 .. pg:3 rn:5 0 za_imm:4 \ LDR 1110000 100 0 000000 .. 000 ..... 0 .... @ldstr STR 1110000 100 1 000000 .. 000 ..... 0 .... @ldstr + +### SME Add Vector to Array + +&adda zad zn pm pn +@adda_32 ........ .. ..... . pm:3 pn:3 zn:5 ... zad:2 &adda +@adda_64 ........ .. ..... . pm:3 pn:3 zn:5 .. zad:3 &adda + +ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 +ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 +ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 +ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ DO_ST(q, _be, MO_128) DO_ST(q, _le, MO_128) #undef DO_ST + +void HELPER(sme_addha_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn[H4(col)]; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addha_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn[col]; + } + } + } + } +} + +void HELPER(sme_addva_s)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 4; + uint64_t *pn = vpn, *pm = vpm; + uint32_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ) { + uint64_t pa = pn[row >> 4]; + do { + if (pa & 1) { + uint32_t zn_row = zn[H4(row)]; + for (col = 0; col < oprsz; ) { + uint64_t pb = pm[col >> 4]; + do { + if (pb & 1) { + zda[tile_vslice_index(row) + H4(col)] += zn_row; + } + pb >>= 4; + } while (++col & 15); + } + } + pa >>= 4; + } while (++row & 15); + } +} + +void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint8_t *pn = vpn, *pm = vpm; + uint64_t *zda = vzda, *zn = vzn; + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t zn_row = zn[row]; + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + zda[tile_vslice_index(row) + col] += zn_row; + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_ldst_r(DisasContext *s, arg_ldstr *a, GenLdStR *fn) TRANS_FEAT(LDR, aa64_sme, do_ldst_r, a, gen_sve_ldr) TRANS_FEAT(STR, aa64_sme, do_ldst_r, a, gen_sve_str) + +static bool do_adda(DisasContext *s, arg_adda *a, MemOp esz, + gen_helper_gvec_4 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, 0); + TCGv_ptr za, zn, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + +TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) +TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) +TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) +TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-25-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 5 +++ target/arm/sme.decode | 9 +++++ target/arm/sme_helper.c | 69 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 32 ++++++++++++++++++ 4 files changed, 115 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addha_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDHA_s 11000000 10 01000 0 ... ... ..... 000 .. @adda_32 ADDVA_s 11000000 10 01000 1 ... ... ..... 000 .. @adda_32 ADDHA_d 11000000 11 01000 0 ... ... ..... 00 ... @adda_64 ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 + +### SME Outer Product + +&op zad zn zm pm pn sub:bool +@op_32 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 .. zad:2 &op +@op_64 ........ ... zm:5 pm:3 pn:3 zn:5 sub:1 . zad:3 &op + +FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ #include "exec/cpu_ldst.h" #include "exec/exec-all.h" #include "qemu/int128.h" +#include "fpu/softfloat.h" #include "vec_internal.h" #include "sve_ldst_internal.h" @@ -XXX,XX +XXX,XX @@ void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn, } } } + +void HELPER(sme_fmopa_s)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) << 31; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. + */ + fpst = *(float_status *)vst; + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ) { + uint16_t pa = pn[H2(row >> 4)]; + do { + if (pa & 1) { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)) ^ neg; + + for (col = 0; col < oprsz; ) { + uint16_t pb = pm[H2(col >> 4)]; + do { + if (pb & 1) { + uint32_t *a = vza_row + H1_4(col); + uint32_t *m = vzm + H1_4(col); + *a = float32_muladd(n, *m, *a, 0, vst); + } + col += 4; + pb >>= 4; + } while (col & 15); + } + } + row += 4; + pa >>= 4; + } while (row & 15); + } +} + +void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + uint64_t neg = (uint64_t)simd_data(desc) << 63; + uint64_t *za = vza, *zn = vzn, *zm = vzm; + uint8_t *pn = vpn, *pm = vpm; + float_status fpst = *(float_status *)vst; + + set_default_nan_mode(true, &fpst); + + for (row = 0; row < oprsz; ++row) { + if (pn[H1(row)] & 1) { + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row] ^ neg; + + for (col = 0; col < oprsz; ++col) { + if (pm[H1(col)] & 1) { + uint64_t *a = &za_row[col]; + *a = float64_muladd(n, zm[col], *a, 0, &fpst); + } + } + } + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDHA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addha_s) TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) + +static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5_ptr *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm, fpst; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + fpst = fpstatus_ptr(FPST_FPCR); + + fn(za, zn, zm, pn, pm, fpst, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + tcg_temp_free_ptr(fpst); + return true; +} + +TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) +TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-26-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 2 ++ target/arm/sme_helper.c | 56 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 30 ++++++++++++++++++++ 4 files changed, 90 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ... @adda_64 FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 + +BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm, void *vpn, } } } + +/* + * Alter PAIR as needed for controlling predicates being false, + * and for NEG on an enabled row element. + */ +static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) +{ + /* + * The pseudocode uses a conditional negate after the conditional zero. + * It is simpler here to unconditionally negate before conditional zero. + */ + pair ^= neg; + if (!(pg & 1)) { + pair &= 0xffff0000u; + } + if (!(pg & 4)) { + pair &= 0x0000ffffu; + } + return pair; +} + +void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = bfdotadd(*a, n, m); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32, gen_helper_sme_addva_s) TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d) TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d) +static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz, + gen_helper_gvec_5 *fn) +{ + int svl = streaming_vec_reg_size(s); + uint32_t desc = simd_desc(svl, svl, a->sub); + TCGv_ptr za, zn, zm, pn, pm; + + if (!sme_smza_enabled_check(s)) { + return true; + } + + /* Sum XZR+zad to find ZAd. */ + za = get_tile_rowcol(s, esz, 31, a->zad, false); + zn = vec_full_reg_ptr(s, a->zn); + zm = vec_full_reg_ptr(s, a->zm); + pn = pred_full_reg_ptr(s, a->pn); + pm = pred_full_reg_ptr(s, a->pm); + + fn(za, zn, zm, pn, pm, tcg_constant_i32(desc)); + + tcg_temp_free_ptr(za); + tcg_temp_free_ptr(zn); + tcg_temp_free_ptr(pn); + tcg_temp_free_ptr(pm); + return true; +} + static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, gen_helper_gvec_5_ptr *fn) { @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) + +/* TODO: FEAT_EBF16 */ +TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-27-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 2 ++ target/arm/sme.decode | 1 + target/arm/sme_helper.c | 74 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 1 + 4 files changed, 78 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(sme_addva_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addha_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_5(sme_addva_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_7(sme_fmopa_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 +FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t neg) return pair; } +static float32 f16_dotadd(float32 sum, uint32_t e1, uint32_t e2, + float_status *s_std, float_status *s_odd) +{ + float64 e1r = float16_to_float64(e1 & 0xffff, true, s_std); + float64 e1c = float16_to_float64(e1 >> 16, true, s_std); + float64 e2r = float16_to_float64(e2 & 0xffff, true, s_std); + float64 e2c = float16_to_float64(e2 >> 16, true, s_std); + float64 t64; + float32 t32; + + /* + * The ARM pseudocode function FPDot performs both multiplies + * and the add with a single rounding operation. Emulate this + * by performing the first multiply in round-to-odd, then doing + * the second multiply as fused multiply-add, and rounding to + * float32 all in one step. + */ + t64 = float64_mul(e1r, e2r, s_odd); + t64 = float64r32_muladd(e1c, e2c, t64, 0, s_std); + + /* This conversion is exact, because we've already rounded. */ + t32 = float64_to_float32(t64, s_std); + + /* The final accumulation step is not fused. */ + return float32_add(sum, t32, s_std); +} + +void HELPER(sme_fmopa_h)(void *vza, void *vzn, void *vzm, void *vpn, + void *vpm, void *vst, uint32_t desc) +{ + intptr_t row, col, oprsz = simd_maxsz(desc); + uint32_t neg = simd_data(desc) * 0x80008000u; + uint16_t *pn = vpn, *pm = vpm; + float_status fpst_odd, fpst_std; + + /* + * Make a copy of float_status because this operation does not + * update the cumulative fp exception status. It also produces + * default nans. Make a second copy with round-to-odd -- see above. + */ + fpst_std = *(float_status *)vst; + set_default_nan_mode(true, &fpst_std); + fpst_odd = fpst_std; + set_float_rounding_mode(float_round_to_odd, &fpst_odd); + + for (row = 0; row < oprsz; ) { + uint16_t prow = pn[H2(row >> 4)]; + do { + void *vza_row = vza + tile_vslice_offset(row); + uint32_t n = *(uint32_t *)(vzn + H1_4(row)); + + n = f16mop_adj_pair(n, prow, neg); + + for (col = 0; col < oprsz; ) { + uint16_t pcol = pm[H2(col >> 4)]; + do { + if (prow & pcol & 0b0101) { + uint32_t *a = vza_row + H1_4(col); + uint32_t m = *(uint32_t *)(vzm + H1_4(col)); + + m = f16mop_adj_pair(m, pcol, 0); + *a = f16_dotadd(*a, n, m, &fpst_std, &fpst_odd); + + col += 4; + pcol >>= 4; + } + } while (col & 15); + } + row += 4; + prow >>= 4; + } while (row & 15); + } +} + void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, void *vpm, uint32_t desc) { diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz, return true; } +TRANS_FEAT(FMOPA_h, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_h) TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s) TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is SMOPA, SUMOPA, USMOPA_s, UMOPA, for both Int8 and Int16. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-28-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sme.h | 16 ++++++++ target/arm/sme.decode | 10 +++++ target/arm/sme_helper.c | 82 ++++++++++++++++++++++++++++++++++++++ target/arm/translate-sme.c | 10 +++++ 4 files changed, 118 insertions(+) diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sme.h +++ b/target/arm/helper-sme.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_smopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_umopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_sumopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_6(sme_usmopa_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, ptr, i32) diff --git a/target/arm/sme.decode b/target/arm/sme.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme.decode +++ b/target/arm/sme.decode @@ -XXX,XX +XXX,XX @@ FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64 BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32 FMOPA_h 10000001 101 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_s 1010000 0 10 0 ..... ... ... ..... . 00 .. @op_32 +SUMOPA_s 1010000 0 10 1 ..... ... ... ..... . 00 .. @op_32 +USMOPA_s 1010000 1 10 0 ..... ... ... ..... . 00 .. @op_32 +UMOPA_s 1010000 1 10 1 ..... ... ... ..... . 00 .. @op_32 + +SMOPA_d 1010000 0 11 0 ..... ... ... ..... . 0 ... @op_64 +SUMOPA_d 1010000 0 11 1 ..... ... ... ..... . 0 ... @op_64 +USMOPA_d 1010000 1 11 0 ..... ... ... ..... . 0 ... @op_64 +UMOPA_d 1010000 1 11 1 ..... ... ... ..... . 0 ... @op_64 diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sme_helper.c +++ b/target/arm/sme_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn, } while (row & 15); } } + +typedef uint64_t IMOPFn(uint64_t, uint64_t, uint64_t, uint8_t, bool); + +static inline void do_imopa(uint64_t *za, uint64_t *zn, uint64_t *zm, + uint8_t *pn, uint8_t *pm, + uint32_t desc, IMOPFn *fn) +{ + intptr_t row, col, oprsz = simd_oprsz(desc) / 8; + bool neg = simd_data(desc); + + for (row = 0; row < oprsz; ++row) { + uint8_t pa = pn[H1(row)]; + uint64_t *za_row = &za[tile_vslice_index(row)]; + uint64_t n = zn[row]; + + for (col = 0; col < oprsz; ++col) { + uint8_t pb = pm[H1(col)]; + uint64_t *a = &za_row[col]; + + *a = fn(n, zm[col], *a, pa & pb, neg); + } + } +} + +#define DEF_IMOP_32(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint32_t sum0 = 0, sum1 = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_b(p); \ + sum0 += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum0 += (NTYPE)(n >> 8) * (MTYPE)(m >> 8); \ + sum0 += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum0 += (NTYPE)(n >> 24) * (MTYPE)(m >> 24); \ + sum1 += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum1 += (NTYPE)(n >> 40) * (MTYPE)(m >> 40); \ + sum1 += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + sum1 += (NTYPE)(n >> 56) * (MTYPE)(m >> 56); \ + if (neg) { \ + sum0 = (uint32_t)a - sum0, sum1 = (uint32_t)(a >> 32) - sum1; \ + } else { \ + sum0 = (uint32_t)a + sum0, sum1 = (uint32_t)(a >> 32) + sum1; \ + } \ + return ((uint64_t)sum1 << 32) | sum0; \ +} + +#define DEF_IMOP_64(NAME, NTYPE, MTYPE) \ +static uint64_t NAME(uint64_t n, uint64_t m, uint64_t a, uint8_t p, bool neg) \ +{ \ + uint64_t sum = 0; \ + /* Apply P to N as a mask, making the inactive elements 0. */ \ + n &= expand_pred_h(p); \ + sum += (NTYPE)(n >> 0) * (MTYPE)(m >> 0); \ + sum += (NTYPE)(n >> 16) * (MTYPE)(m >> 16); \ + sum += (NTYPE)(n >> 32) * (MTYPE)(m >> 32); \ + sum += (NTYPE)(n >> 48) * (MTYPE)(m >> 48); \ + return neg ? a - sum : a + sum; \ +} + +DEF_IMOP_32(smopa_s, int8_t, int8_t) +DEF_IMOP_32(umopa_s, uint8_t, uint8_t) +DEF_IMOP_32(sumopa_s, int8_t, uint8_t) +DEF_IMOP_32(usmopa_s, uint8_t, int8_t) + +DEF_IMOP_64(smopa_d, int16_t, int16_t) +DEF_IMOP_64(umopa_d, uint16_t, uint16_t) +DEF_IMOP_64(sumopa_d, int16_t, uint16_t) +DEF_IMOP_64(usmopa_d, uint16_t, int16_t) + +#define DEF_IMOPH(NAME) \ + void HELPER(sme_##NAME)(void *vza, void *vzn, void *vzm, void *vpn, \ + void *vpm, uint32_t desc) \ + { do_imopa(vza, vzn, vzm, vpn, vpm, desc, NAME); } + +DEF_IMOPH(smopa_s) +DEF_IMOPH(umopa_s) +DEF_IMOPH(sumopa_s) +DEF_IMOPH(usmopa_s) +DEF_IMOPH(smopa_d) +DEF_IMOPH(umopa_d) +DEF_IMOPH(sumopa_d) +DEF_IMOPH(usmopa_d) diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sme.c +++ b/target/arm/translate-sme.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_f /* TODO: FEAT_EBF16 */ TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa) + +TRANS_FEAT(SMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_smopa_s) +TRANS_FEAT(UMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_umopa_s) +TRANS_FEAT(SUMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_sumopa_s) +TRANS_FEAT(USMOPA_s, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_usmopa_s) + +TRANS_FEAT(SMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_smopa_d) +TRANS_FEAT(UMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_umopa_d) +TRANS_FEAT(SUMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_sumopa_d) +TRANS_FEAT(USMOPA_d, aa64_sme_i16i64, do_outprod, a, MO_64, gen_helper_sme_usmopa_d) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-29-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/sve.decode | 20 +++++++++++++ target/arm/translate-sve.c | 57 ++++++++++++++++++++++++++++++++++++++ 2 files changed, 77 insertions(+) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ BFMLALT_zzxw 01100100 11 1 ..... 0100.1 ..... ..... @rrxr_3a esz=2 ### SVE2 floating-point bfloat16 dot-product (indexed) BFDOT_zzxz 01100100 01 1 ..... 010000 ..... ..... @rrxr_2 esz=2 + +### SVE broadcast predicate element + +&psel esz pd pn pm rv imm +%psel_rv 16:2 !function=plus_12 +%psel_imm_b 22:2 19:2 +%psel_imm_h 22:2 20:1 +%psel_imm_s 22:2 +%psel_imm_d 23:1 +@psel ........ .. . ... .. .. pn:4 . pm:4 . pd:4 \ + &psel rv=%psel_rv + +PSEL 00100101 .. 1 ..1 .. 01 .... 0 .... 0 .... \ + @psel esz=0 imm=%psel_imm_b +PSEL 00100101 .. 1 .10 .. 01 .... 0 .... 0 .... \ + @psel esz=1 imm=%psel_imm_h +PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ + @psel esz=2 imm=%psel_imm_s +PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ + @psel esz=3 imm=%psel_imm_d diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool do_BFMLAL_zzxw(DisasContext *s, arg_rrxr_esz *a, bool sel) TRANS_FEAT(BFMLALB_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, false) TRANS_FEAT(BFMLALT_zzxw, aa64_sve_bf16, do_BFMLAL_zzxw, a, true) + +static bool trans_PSEL(DisasContext *s, arg_psel *a) +{ + int vl = vec_full_reg_size(s); + int pl = pred_gvec_reg_size(s); + int elements = vl >> a->esz; + TCGv_i64 tmp, didx, dbit; + TCGv_ptr ptr; + + if (!dc_isar_feature(aa64_sme, s)) { + return false; + } + if (!sve_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i64(); + dbit = tcg_temp_new_i64(); + didx = tcg_temp_new_i64(); + ptr = tcg_temp_new_ptr(); + + /* Compute the predicate element. */ + tcg_gen_addi_i64(tmp, cpu_reg(s, a->rv), a->imm); + if (is_power_of_2(elements)) { + tcg_gen_andi_i64(tmp, tmp, elements - 1); + } else { + tcg_gen_remu_i64(tmp, tmp, tcg_constant_i64(elements)); + } + + /* Extract the predicate byte and bit indices. */ + tcg_gen_shli_i64(tmp, tmp, a->esz); + tcg_gen_andi_i64(dbit, tmp, 7); + tcg_gen_shri_i64(didx, tmp, 3); + if (HOST_BIG_ENDIAN) { + tcg_gen_xori_i64(didx, didx, 7); + } + + /* Load the predicate word. */ + tcg_gen_trunc_i64_ptr(ptr, didx); + tcg_gen_add_ptr(ptr, ptr, cpu_env); + tcg_gen_ld8u_i64(tmp, ptr, pred_full_reg_offset(s, a->pm)); + + /* Extract the predicate bit and replicate to MO_64. */ + tcg_gen_shr_i64(tmp, tmp, dbit); + tcg_gen_andi_i64(tmp, tmp, 1); + tcg_gen_neg_i64(tmp, tmp); + + /* Apply to either copy the source, or write zeros. */ + tcg_gen_gvec_ands(MO_64, pred_full_reg_offset(s, a->pd), + pred_full_reg_offset(s, a->pn), tmp, pl, pl); + + tcg_temp_free_i64(tmp); + tcg_temp_free_i64(dbit); + tcg_temp_free_i64(didx); + tcg_temp_free_ptr(ptr); + return true; +} -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-30-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper-sve.h | 2 ++ target/arm/sve.decode | 1 + target/arm/sve_helper.c | 16 ++++++++++++++++ target/arm/translate-sve.c | 2 ++ 4 files changed, 21 insertions(+) diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper-sve.h +++ b/target/arm/helper-sve.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_4(sve_revh_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_revw_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_4(sme_revd_q, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) + DEF_HELPER_FLAGS_4(sve_rbit_b, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) DEF_HELPER_FLAGS_4(sve_rbit_s, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32) diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ REVB 00000101 .. 1001 00 100 ... ..... ..... @rd_pg_rn REVH 00000101 .. 1001 01 100 ... ..... ..... @rd_pg_rn REVW 00000101 .. 1001 10 100 ... ..... ..... @rd_pg_rn RBIT 00000101 .. 1001 11 100 ... ..... ..... @rd_pg_rn +REVD 00000101 00 1011 10 100 ... ..... ..... @rd_pg_rn_e0 # SVE vector splice (predicated, destructive) SPLICE 00000101 .. 101 100 100 ... ..... ..... @rdn_pg_rm diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve_helper.c +++ b/target/arm/sve_helper.c @@ -XXX,XX +XXX,XX @@ DO_ZPZ_D(sve_revh_d, uint64_t, hswap64) DO_ZPZ_D(sve_revw_d, uint64_t, wswap64) +void HELPER(sme_revd_q)(void *vd, void *vn, void *vg, uint32_t desc) +{ + intptr_t i, opr_sz = simd_oprsz(desc) / 8; + uint64_t *d = vd, *n = vn; + uint8_t *pg = vg; + + for (i = 0; i < opr_sz; i += 2) { + if (pg[H1(i)] & 1) { + uint64_t n0 = n[i + 0]; + uint64_t n1 = n[i + 1]; + d[i + 0] = n1; + d[i + 1] = n0; + } + } +} + DO_ZPZ(sve_rbit_b, uint8_t, H1, revbit8) DO_ZPZ(sve_rbit_h, uint16_t, H1_2, revbit16) DO_ZPZ(sve_rbit_s, uint32_t, H1_4, revbit32) diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ TRANS_FEAT(REVH, aa64_sve, gen_gvec_ool_arg_zpz, revh_fns[a->esz], a, 0) TRANS_FEAT(REVW, aa64_sve, gen_gvec_ool_arg_zpz, a->esz == 3 ? gen_helper_sve_revw_d : NULL, a, 0) +TRANS_FEAT(REVD, aa64_sme, gen_gvec_ool_arg_zpz, gen_helper_sme_revd_q, a, 0) + TRANS_FEAT(SPLICE, aa64_sve, gen_gvec_ool_arg_zpzz, gen_helper_sve_splice, a, a->esz) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> This is an SVE instruction that operates using the SVE vector length but that it is present only if SME is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-31-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.h | 18 +++++++ target/arm/sve.decode | 5 ++ target/arm/translate-sve.c | 102 +++++++++++++++++++++++++++++++++++++ target/arm/vec_helper.c | 24 +++++++++ 4 files changed, 149 insertions(+) diff --git a/target/arm/helper.h b/target/arm/helper.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.h +++ b/target/arm/helper.h @@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_6(gvec_bfmlal, TCG_CALL_NO_RWG, DEF_HELPER_FLAGS_6(gvec_bfmlal_idx, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_sclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + +DEF_HELPER_FLAGS_5(gvec_uclamp_b, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_h, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_s, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) +DEF_HELPER_FLAGS_5(gvec_uclamp_d, TCG_CALL_NO_RWG, + void, ptr, ptr, ptr, ptr, i32) + #ifdef TARGET_AARCH64 #include "helper-a64.h" #include "helper-sve.h" diff --git a/target/arm/sve.decode b/target/arm/sve.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/sve.decode +++ b/target/arm/sve.decode @@ -XXX,XX +XXX,XX @@ PSEL 00100101 .. 1 100 .. 01 .... 0 .... 0 .... \ @psel esz=2 imm=%psel_imm_s PSEL 00100101 .1 1 000 .. 01 .... 0 .... 0 .... \ @psel esz=3 imm=%psel_imm_d + +### SVE clamp + +SCLAMP 01000100 .. 0 ..... 110000 ..... ..... @rda_rn_rm +UCLAMP 01000100 .. 0 ..... 110001 ..... ..... @rda_rn_rm diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-sve.c +++ b/target/arm/translate-sve.c @@ -XXX,XX +XXX,XX @@ static bool trans_PSEL(DisasContext *s, arg_psel *a) tcg_temp_free_ptr(ptr); return true; } + +static void gen_sclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_smax_i32(d, a, n); + tcg_gen_smin_i32(d, d, m); +} + +static void gen_sclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_smax_i64(d, a, n); + tcg_gen_smin_i64(d, d, m); +} + +static void gen_sclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_smax_vec(vece, d, a, n); + tcg_gen_smin_vec(vece, d, d, m); +} + +static void gen_sclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_smin_vec, INDEX_op_smax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_sclamp_i32, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_sclamp_i64, + .fniv = gen_sclamp_vec, + .fno = gen_helper_gvec_sclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(SCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_sclamp, a) + +static void gen_uclamp_i32(TCGv_i32 d, TCGv_i32 n, TCGv_i32 m, TCGv_i32 a) +{ + tcg_gen_umax_i32(d, a, n); + tcg_gen_umin_i32(d, d, m); +} + +static void gen_uclamp_i64(TCGv_i64 d, TCGv_i64 n, TCGv_i64 m, TCGv_i64 a) +{ + tcg_gen_umax_i64(d, a, n); + tcg_gen_umin_i64(d, d, m); +} + +static void gen_uclamp_vec(unsigned vece, TCGv_vec d, TCGv_vec n, + TCGv_vec m, TCGv_vec a) +{ + tcg_gen_umax_vec(vece, d, a, n); + tcg_gen_umin_vec(vece, d, d, m); +} + +static void gen_uclamp(unsigned vece, uint32_t d, uint32_t n, uint32_t m, + uint32_t a, uint32_t oprsz, uint32_t maxsz) +{ + static const TCGOpcode vecop[] = { + INDEX_op_umin_vec, INDEX_op_umax_vec, 0 + }; + static const GVecGen4 ops[4] = { + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_b, + .opt_opc = vecop, + .vece = MO_8 }, + { .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_h, + .opt_opc = vecop, + .vece = MO_16 }, + { .fni4 = gen_uclamp_i32, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_s, + .opt_opc = vecop, + .vece = MO_32 }, + { .fni8 = gen_uclamp_i64, + .fniv = gen_uclamp_vec, + .fno = gen_helper_gvec_uclamp_d, + .opt_opc = vecop, + .vece = MO_64, + .prefer_i64 = TCG_TARGET_REG_BITS == 64 } + }; + tcg_gen_gvec_4(d, n, m, a, oprsz, maxsz, &ops[vece]); +} + +TRANS_FEAT(UCLAMP, aa64_sme, gen_gvec_fn_arg_zzzz, gen_uclamp, a) diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -XXX,XX +XXX,XX @@ void HELPER(gvec_bfmlal_idx)(void *vd, void *vn, void *vm, } clear_tail(d, opr_sz, simd_maxsz(desc)); } + +#define DO_CLAMP(NAME, TYPE) \ +void HELPER(NAME)(void *d, void *n, void *m, void *a, uint32_t desc) \ +{ \ + intptr_t i, opr_sz = simd_oprsz(desc); \ + for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ + TYPE aa = *(TYPE *)(a + i); \ + TYPE nn = *(TYPE *)(n + i); \ + TYPE mm = *(TYPE *)(m + i); \ + TYPE dd = MIN(MAX(aa, nn), mm); \ + *(TYPE *)(d + i) = dd; \ + } \ + clear_tail(d, opr_sz, simd_maxsz(desc)); \ +} + +DO_CLAMP(gvec_sclamp_b, int8_t) +DO_CLAMP(gvec_sclamp_h, int16_t) +DO_CLAMP(gvec_sclamp_s, int32_t) +DO_CLAMP(gvec_sclamp_d, int64_t) + +DO_CLAMP(gvec_uclamp_b, uint8_t) +DO_CLAMP(gvec_uclamp_h, uint16_t) +DO_CLAMP(gvec_uclamp_s, uint32_t) +DO_CLAMP(gvec_uclamp_d, uint64_t) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> We can handle both exception entry and exception return by hooking into aarch64_sve_change_el. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-32-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, return; } + old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; + new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; + + /* + * Both AArch64.TakeException and AArch64.ExceptionReturn + * invoke ResetSVEState when taking an exception from, or + * returning to, AArch32 state when PSTATE.SM is enabled. + */ + if (old_a64 != new_a64 && FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + return; + } + /* * DDI0584A.d sec 3.2: "If SVE instructions are disabled or trapped * at ELx, or not available because the EL is in AArch32 state, then @@ -XXX,XX +XXX,XX @@ void aarch64_sve_change_el(CPUARMState *env, int old_el, * we already have the correct register contents when encountering the * vq0->vq0 transition between EL0->EL1. */ - old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64; old_len = (old_a64 && !sve_exception_el(env, old_el) ? sve_vqm1_for_el(env, old_el) : 0); - new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64; new_len = (new_a64 && !sve_exception_el(env, new_el) ? sve_vqm1_for_el(env, new_el) : 0); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Note that SME remains effectively disabled for user-only, because we do not yet set CPACR_EL1.SMEN. This needs to wait until the kernel ABI is implemented. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-33-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- docs/system/arm/emulation.rst | 4 ++++ target/arm/cpu64.c | 11 +++++++++++ 2 files changed, 15 insertions(+) diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/system/arm/emulation.rst +++ b/docs/system/arm/emulation.rst @@ -XXX,XX +XXX,XX @@ the following architecture extensions: - FEAT_SHA512 (Advanced SIMD SHA512 instructions) - FEAT_SM3 (Advanced SIMD SM3 instructions) - FEAT_SM4 (Advanced SIMD SM4 instructions) +- FEAT_SME (Scalable Matrix Extension) +- FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode) +- FEAT_SME_F64F64 (Double-precision floating-point outer product instructions) +- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions) - FEAT_SPECRES (Speculation restriction instructions) - FEAT_SSBS (Speculative Store Bypass Safe) - FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain) diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) */ t = FIELD_DP64(t, ID_AA64PFR1, MTE, 3); /* FEAT_MTE3 */ t = FIELD_DP64(t, ID_AA64PFR1, RAS_FRAC, 0); /* FEAT_RASv1p1 + FEAT_DoubleFault */ + t = FIELD_DP64(t, ID_AA64PFR1, SME, 1); /* FEAT_SME */ t = FIELD_DP64(t, ID_AA64PFR1, CSV2_FRAC, 0); /* FEAT_CSV2_2 */ cpu->isar.id_aa64pfr1 = t; @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) t = FIELD_DP64(t, ID_AA64DFR0, PMUVER, 5); /* FEAT_PMUv3p4 */ cpu->isar.id_aa64dfr0 = t; + t = cpu->isar.id_aa64smfr0; + t = FIELD_DP64(t, ID_AA64SMFR0, F32F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, B16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F16F32, 1); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, I8I32, 0xf); /* FEAT_SME */ + t = FIELD_DP64(t, ID_AA64SMFR0, F64F64, 1); /* FEAT_SME_F64F64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, I16I64, 0xf); /* FEAT_SME_I16I64 */ + t = FIELD_DP64(t, ID_AA64SMFR0, FA64, 1); /* FEAT_SME_FA64 */ + cpu->isar.id_aa64smfr0 = t; + /* Replicate the same data to the 32-bit id registers. */ aa32_max_features(cpu); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-34-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_cpu.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/target_cpu.h b/linux-user/aarch64/target_cpu.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_cpu.h +++ b/linux-user/aarch64/target_cpu.h @@ -XXX,XX +XXX,XX @@ static inline void cpu_clone_regs_parent(CPUARMState *env, unsigned flags) static inline void cpu_set_tls(CPUARMState *env, target_ulong newtls) { - /* Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is + /* + * Note that AArch64 Linux keeps the TLS pointer in TPIDR; this is * different from AArch32 Linux, which uses TPIDRRO. */ env->cp15.tpidr_el[0] = newtls; + /* TPIDR2_EL0 is cleared with CLONE_SETTLS. */ + env->cp15.tpidr2_el0 = 0; } static inline abi_ulong get_sp_from_cpustate(CPUARMState *state) -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-35-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/cpu_loop.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/linux-user/aarch64/cpu_loop.c b/linux-user/aarch64/cpu_loop.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/cpu_loop.c +++ b/linux-user/aarch64/cpu_loop.c @@ -XXX,XX +XXX,XX @@ void cpu_loop(CPUARMState *env) switch (trapnr) { case EXCP_SWI: + /* + * On syscall, PSTATE.ZA is preserved, along with the ZA matrix. + * PSTATE.SM is cleared, per SMSTOP, which does ResetSVEState. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + env->svcr = FIELD_DP64(env->svcr, SVCR, SM, 0); + arm_rebuild_hflags(env); + arm_reset_sve_state(env); + } ret = do_syscall(env, env->xregs[8], env->xregs[0], -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Make sure to zero the currently reserved fields. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-36-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_extra_context { struct target_sve_context { struct target_aarch64_ctx head; uint16_t vl; - uint16_t reserved[3]; + uint16_t flags; + uint16_t reserved[2]; /* The actual SVE data immediately follows. It is laid out * according to TARGET_SVE_SIG_{Z,P}REG_OFFSET, based off of * the original struct pointer. @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_CONTEXT_SIZE(VQ) \ (TARGET_SVE_SIG_PREG_OFFSET(VQ, 17)) +#define TARGET_SVE_SIG_FLAG_SM 1 + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, { int i, j; + memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); __put_user(size, &sve->head.size); __put_user(vq * TARGET_SVE_VQ_BYTES, &sve->vl); + if (FIELD_EX64(env->svcr, SVCR, SM)) { + __put_user(TARGET_SVE_SIG_FLAG_SM, &sve->flags); + } /* Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian store -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Fold the return value setting into the goto, so each point of failure need not do both. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-37-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 26 +++++++++++--------------- 1 file changed, 11 insertions(+), 15 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - bool err = false; int vq = 0, sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, switch (magic) { case 0: if (size != 0) { - err = true; - goto exit; + goto err; } if (used_extra) { ctx = NULL; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, case TARGET_FPSIMD_MAGIC: if (fpsimd || size != sizeof(struct target_fpsimd_context)) { - err = true; - goto exit; + goto err; } fpsimd = (struct target_fpsimd_context *)ctx; break; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; } } - err = true; - goto exit; + goto err; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { - err = true; - goto exit; + goto err; } __get_user(extra_datap, &((struct target_extra_context *)ctx)->datap); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, /* Unknown record -- we certainly didn't generate it. * Did we in fact get out of sync? */ - err = true; - goto exit; + goto err; } ctx = (void *)ctx + size; } @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (fpsimd) { target_restore_fpsimd_record(env, fpsimd); } else { - err = true; + goto err; } /* SVE data, if present, overwrites FPSIMD data. */ if (sve) { target_restore_sve_record(env, sve, vq); } - - exit: unlock_user(extra, extra_datap, 0); - return err; + return 0; + + err: + unlock_user(extra, extra_datap, 0); + return 1; } static abi_ulong get_sigframe(struct target_sigaction *ka, -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> In parse_user_sigframe, the kernel rejects duplicate sve records, or records that are smaller than the header. We were silently allowing these cases to pass, dropping the record. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-38-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, break; case TARGET_SVE_MAGIC: + if (sve || size < sizeof(struct target_sve_context)) { + goto err; + } if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { vq = sve_vq(env); sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (!sve && size == sve_size) { + if (size == sve_size) { sve = (struct target_sve_context *)ctx; break; } -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-39-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, __get_user(extra_size, &((struct target_extra_context *)ctx)->size); extra = lock_user(VERIFY_READ, extra_datap, extra_size, 0); + if (!extra) { + return 1; + } break; default: -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Move the checks out of the parsing loop and into the restore function. This more closely mirrors the code structure in the kernel, and is slightly clearer. Reject rather than silently skip incorrect VL and SVE record sizes, bringing our checks in to line with those the kernel does. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-40-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 51 +++++++++++++++++++++++++------------ 1 file changed, 35 insertions(+), 16 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, } } -static void target_restore_sve_record(CPUARMState *env, - struct target_sve_context *sve, int vq) +static bool target_restore_sve_record(CPUARMState *env, + struct target_sve_context *sve, + int size) { - int i, j; + int i, j, vl, vq; - /* Note that SVE regs are stored as a byte stream, with each byte element + if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + return false; + } + + __get_user(vl, &sve->vl); + vq = sve_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.SM. */ + if (size <= sizeof(*sve)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_SVE_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + /* + * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load * of our 64-bit hunks. */ @@ -XXX,XX +XXX,XX @@ static void target_restore_sve_record(CPUARMState *env, } } } + return true; } static int target_restore_sigframe(CPUARMState *env, @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, struct target_sve_context *sve = NULL; uint64_t extra_datap = 0; bool used_extra = false; - int vq = 0, sve_size = 0; + int sve_size = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, if (sve || size < sizeof(struct target_sve_context)) { goto err; } - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); - if (size == sve_size) { - sve = (struct target_sve_context *)ctx; - break; - } - } - goto err; + sve = (struct target_sve_context *)ctx; + sve_size = size; + break; case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve) { - target_restore_sve_record(env, sve, vq); + if (sve && !target_restore_sve_record(env, sve, sve_size)) { + goto err; } unlock_user(extra, extra_datap, 0); return 0; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Set the SM bit in the SVE record on signal delivery, create the ZA record. Restore SM and ZA state according to the records present on return. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-41-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/signal.c | 167 +++++++++++++++++++++++++++++++++--- 1 file changed, 154 insertions(+), 13 deletions(-) diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -XXX,XX +XXX,XX @@ struct target_sve_context { #define TARGET_SVE_SIG_FLAG_SM 1 +#define TARGET_ZA_MAGIC 0x54366345 + +struct target_za_context { + struct target_aarch64_ctx head; + uint16_t vl; + uint16_t reserved[3]; + /* The actual ZA data immediately follows. */ +}; + +#define TARGET_ZA_SIG_REGS_OFFSET \ + QEMU_ALIGN_UP(sizeof(struct target_za_context), TARGET_SVE_VQ_BYTES) +#define TARGET_ZA_SIG_ZAV_OFFSET(VQ, N) \ + (TARGET_ZA_SIG_REGS_OFFSET + (VQ) * TARGET_SVE_VQ_BYTES * (N)) +#define TARGET_ZA_SIG_CONTEXT_SIZE(VQ) \ + TARGET_ZA_SIG_ZAV_OFFSET(VQ, VQ * TARGET_SVE_VQ_BYTES) + struct target_rt_sigframe { struct target_siginfo info; struct target_ucontext uc; @@ -XXX,XX +XXX,XX @@ static void target_setup_end_record(struct target_aarch64_ctx *end) } static void target_setup_sve_record(struct target_sve_context *sve, - CPUARMState *env, int vq, int size) + CPUARMState *env, int size) { - int i, j; + int i, j, vq = sve_vq(env); memset(sve, 0, sizeof(*sve)); __put_user(TARGET_SVE_MAGIC, &sve->head.magic); @@ -XXX,XX +XXX,XX @@ static void target_setup_sve_record(struct target_sve_context *sve, } } +static void target_setup_za_record(struct target_za_context *za, + CPUARMState *env, int size) +{ + int vq = sme_vq(env); + int vl = vq * TARGET_SVE_VQ_BYTES; + int i, j; + + memset(za, 0, sizeof(*za)); + __put_user(TARGET_ZA_MAGIC, &za->head.magic); + __put_user(size, &za->head.size); + __put_user(vl, &za->vl); + + if (size == TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return; + } + assert(size == TARGET_ZA_SIG_CONTEXT_SIZE(vq)); + + /* + * Note that ZA vectors are stored as a byte stream, + * with each byte element at a subsequent address. + */ + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __put_user_e(env->zarray[i].d[j], z + j, le); + } + } +} + static void target_restore_general_frame(CPUARMState *env, struct target_rt_sigframe *sf) { @@ -XXX,XX +XXX,XX @@ static void target_restore_fpsimd_record(CPUARMState *env, static bool target_restore_sve_record(CPUARMState *env, struct target_sve_context *sve, - int size) + int size, int *svcr) { - int i, j, vl, vq; + int i, j, vl, vq, flags; + bool sm; - if (!cpu_isar_feature(aa64_sve, env_archcpu(env))) { + __get_user(vl, &sve->vl); + __get_user(flags, &sve->flags); + + sm = flags & TARGET_SVE_SIG_FLAG_SM; + + /* The cpu must support Streaming or Non-streaming SVE. */ + if (sm + ? !cpu_isar_feature(aa64_sme, env_archcpu(env)) + : !cpu_isar_feature(aa64_sve, env_archcpu(env))) { return false; } - __get_user(vl, &sve->vl); - vq = sve_vq(env); + /* + * Note that we cannot use sve_vq() because that depends on the + * current setting of PSTATE.SM, not the state to be restored. + */ + vq = sve_vqm1_for_el_sm(env, 0, sm) + 1; /* Reject mismatched VL. */ if (vl != vq * TARGET_SVE_VQ_BYTES) { @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return false; } + *svcr = FIELD_DP64(*svcr, SVCR, SM, sm); + /* * Note that SVE regs are stored as a byte stream, with each byte element * at a subsequent address. This corresponds to a little-endian load @@ -XXX,XX +XXX,XX @@ static bool target_restore_sve_record(CPUARMState *env, return true; } +static bool target_restore_za_record(CPUARMState *env, + struct target_za_context *za, + int size, int *svcr) +{ + int i, j, vl, vq; + + if (!cpu_isar_feature(aa64_sme, env_archcpu(env))) { + return false; + } + + __get_user(vl, &za->vl); + vq = sme_vq(env); + + /* Reject mismatched VL. */ + if (vl != vq * TARGET_SVE_VQ_BYTES) { + return false; + } + + /* Accept empty record -- used to clear PSTATE.ZA. */ + if (size <= TARGET_ZA_SIG_CONTEXT_SIZE(0)) { + return true; + } + + /* Reject non-empty but incomplete record. */ + if (size < TARGET_ZA_SIG_CONTEXT_SIZE(vq)) { + return false; + } + + *svcr = FIELD_DP64(*svcr, SVCR, ZA, 1); + + for (i = 0; i < vl; ++i) { + uint64_t *z = (void *)za + TARGET_ZA_SIG_ZAV_OFFSET(vq, i); + for (j = 0; j < vq * 2; ++j) { + __get_user_e(env->zarray[i].d[j], z + j, le); + } + } + return true; +} + static int target_restore_sigframe(CPUARMState *env, struct target_rt_sigframe *sf) { struct target_aarch64_ctx *ctx, *extra = NULL; struct target_fpsimd_context *fpsimd = NULL; struct target_sve_context *sve = NULL; + struct target_za_context *za = NULL; uint64_t extra_datap = 0; bool used_extra = false; int sve_size = 0; + int za_size = 0; + int svcr = 0; target_restore_general_frame(env, sf); @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, sve_size = size; break; + case TARGET_ZA_MAGIC: + if (za || size < sizeof(struct target_za_context)) { + goto err; + } + za = (struct target_za_context *)ctx; + za_size = size; + break; + case TARGET_EXTRA_MAGIC: if (extra || size != sizeof(struct target_extra_context)) { goto err; @@ -XXX,XX +XXX,XX @@ static int target_restore_sigframe(CPUARMState *env, } /* SVE data, if present, overwrites FPSIMD data. */ - if (sve && !target_restore_sve_record(env, sve, sve_size)) { + if (sve && !target_restore_sve_record(env, sve, sve_size, &svcr)) { goto err; } + if (za && !target_restore_za_record(env, za, za_size, &svcr)) { + goto err; + } + if (env->svcr != svcr) { + env->svcr = svcr; + arm_rebuild_hflags(env); + } unlock_user(extra, extra_datap, 0); return 0; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, .total_size = offsetof(struct target_rt_sigframe, uc.tuc_mcontext.__reserved), }; - int fpsimd_ofs, fr_ofs, sve_ofs = 0, vq = 0, sve_size = 0; + int fpsimd_ofs, fr_ofs, sve_ofs = 0, za_ofs = 0; + int sve_size = 0, za_size = 0; struct target_rt_sigframe *frame; struct target_rt_frame_record *fr; abi_ulong frame_addr, return_addr; @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, &layout); /* SVE state needs saving only if it exists. */ - if (cpu_isar_feature(aa64_sve, env_archcpu(env))) { - vq = sve_vq(env); - sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(vq), 16); + if (cpu_isar_feature(aa64_sve, env_archcpu(env)) || + cpu_isar_feature(aa64_sme, env_archcpu(env))) { + sve_size = QEMU_ALIGN_UP(TARGET_SVE_SIG_CONTEXT_SIZE(sve_vq(env)), 16); sve_ofs = alloc_sigframe_space(sve_size, &layout); } + if (cpu_isar_feature(aa64_sme, env_archcpu(env))) { + /* ZA state needs saving only if it is enabled. */ + if (FIELD_EX64(env->svcr, SVCR, ZA)) { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(sme_vq(env)); + } else { + za_size = TARGET_ZA_SIG_CONTEXT_SIZE(0); + } + za_ofs = alloc_sigframe_space(za_size, &layout); + } if (layout.extra_ofs) { /* Reserve space for the extra end marker. The standard end marker @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, target_setup_end_record((void *)frame + layout.extra_end_ofs); } if (sve_ofs) { - target_setup_sve_record((void *)frame + sve_ofs, env, vq, sve_size); + target_setup_sve_record((void *)frame + sve_ofs, env, sve_size); + } + if (za_ofs) { + target_setup_za_record((void *)frame + za_ofs, env, za_size); } /* Set up the stack frame for unwinding. */ @@ -XXX,XX +XXX,XX @@ static void target_setup_frame(int usig, struct target_sigaction *ka, env->btype = 2; } + /* + * Invoke the signal handler with both SM and ZA disabled. + * When clearing SM, ResetSVEState, per SMSTOP. + */ + if (FIELD_EX64(env->svcr, SVCR, SM)) { + arm_reset_sve_state(env); + } + if (env->svcr) { + env->svcr = 0; + arm_rebuild_hflags(env); + } + if (info) { tswap_siginfo(&frame->info, info); env->xregs[1] = frame_addr + offsetof(struct target_rt_sigframe, info); -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Add "sve" to the sve prctl functions, to distinguish them from the coming "sme" prctls with similar names. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-42-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 8 ++++---- linux-user/syscall.c | 12 ++++++------ 2 files changed, 10 insertions(+), 10 deletions(-) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ #ifndef AARCH64_TARGET_PRCTL_H #define AARCH64_TARGET_PRCTL_H -static abi_long do_prctl_get_vl(CPUArchState *env) +static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_get_vl(CPUArchState *env) } return -TARGET_EINVAL; } -#define do_prctl_get_vl do_prctl_get_vl +#define do_prctl_sve_get_vl do_prctl_sve_get_vl -static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) +static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) { /* * We cannot support either PR_SVE_SET_VL_ONEXEC or PR_SVE_VL_INHERIT. @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_set_vl(CPUArchState *env, abi_long arg2) } return -TARGET_EINVAL; } -#define do_prctl_set_vl do_prctl_set_vl +#define do_prctl_sve_set_vl do_prctl_sve_set_vl static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_fp_mode #define do_prctl_set_fp_mode do_prctl_inval1 #endif -#ifndef do_prctl_get_vl -#define do_prctl_get_vl do_prctl_inval0 +#ifndef do_prctl_sve_get_vl +#define do_prctl_sve_get_vl do_prctl_inval0 #endif -#ifndef do_prctl_set_vl -#define do_prctl_set_vl do_prctl_inval1 +#ifndef do_prctl_sve_set_vl +#define do_prctl_sve_set_vl do_prctl_inval1 #endif #ifndef do_prctl_reset_keys #define do_prctl_reset_keys do_prctl_inval1 @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, case PR_SET_FP_MODE: return do_prctl_set_fp_mode(env, arg2); case PR_SVE_GET_VL: - return do_prctl_get_vl(env); + return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: - return do_prctl_set_vl(env, arg2); + return do_prctl_sve_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> These prctl set the Streaming SVE vector length, which may be completely different from the Normal SVE vector length. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-43-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/aarch64/target_prctl.h | 54 +++++++++++++++++++++++++++++++ linux-user/syscall.c | 16 +++++++++ 2 files changed, 70 insertions(+) diff --git a/linux-user/aarch64/target_prctl.h b/linux-user/aarch64/target_prctl.h index XXXXXXX..XXXXXXX 100644 --- a/linux-user/aarch64/target_prctl.h +++ b/linux-user/aarch64/target_prctl.h @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_get_vl(CPUArchState *env) { ARMCPU *cpu = env_archcpu(env); if (cpu_isar_feature(aa64_sve, cpu)) { + /* PSTATE.SM is always unset on syscall entry. */ return sve_vq(env) * 16; } return -TARGET_EINVAL; @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { uint32_t vq, old_vq; + /* PSTATE.SM is always unset on syscall entry. */ old_vq = sve_vq(env); /* @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_sve_set_vl(CPUArchState *env, abi_long arg2) } #define do_prctl_sve_set_vl do_prctl_sve_set_vl +static abi_long do_prctl_sme_get_vl(CPUArchState *env) +{ + ARMCPU *cpu = env_archcpu(env); + if (cpu_isar_feature(aa64_sme, cpu)) { + return sme_vq(env) * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_get_vl do_prctl_sme_get_vl + +static abi_long do_prctl_sme_set_vl(CPUArchState *env, abi_long arg2) +{ + /* + * We cannot support either PR_SME_SET_VL_ONEXEC or PR_SME_VL_INHERIT. + * Note the kernel definition of sve_vl_valid allows for VQ=512, + * i.e. VL=8192, even though the architectural maximum is VQ=16. + */ + if (cpu_isar_feature(aa64_sme, env_archcpu(env)) + && arg2 >= 0 && arg2 <= 512 * 16 && !(arg2 & 15)) { + int vq, old_vq; + + old_vq = sme_vq(env); + + /* + * Bound the value of vq, so that we know that it fits into + * the 4-bit field in SMCR_EL1. Because PSTATE.SM is cleared + * on syscall entry, we are not modifying the current SVE + * vector length. + */ + vq = MAX(arg2 / 16, 1); + vq = MIN(vq, 16); + env->vfp.smcr_el[1] = + FIELD_DP64(env->vfp.smcr_el[1], SMCR, LEN, vq - 1); + + /* Delay rebuilding hflags until we know if ZA must change. */ + vq = sve_vqm1_for_el_sm(env, 0, true) + 1; + + if (vq != old_vq) { + /* + * PSTATE.ZA state is cleared on any change to SVL. + * We need not call arm_rebuild_hflags because PSTATE.SM was + * cleared on syscall entry, so this hasn't changed VL. + */ + env->svcr = FIELD_DP64(env->svcr, SVCR, ZA, 0); + arm_rebuild_hflags(env); + } + return vq * 16; + } + return -TARGET_EINVAL; +} +#define do_prctl_sme_set_vl do_prctl_sme_set_vl + static abi_long do_prctl_reset_keys(CPUArchState *env, abi_long arg2) { ARMCPU *cpu = env_archcpu(env); diff --git a/linux-user/syscall.c b/linux-user/syscall.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -XXX,XX +XXX,XX @@ abi_long do_arch_prctl(CPUX86State *env, int code, abi_ulong addr) #ifndef PR_SET_SYSCALL_USER_DISPATCH # define PR_SET_SYSCALL_USER_DISPATCH 59 #endif +#ifndef PR_SME_SET_VL +# define PR_SME_SET_VL 63 +# define PR_SME_GET_VL 64 +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) +#endif #include "target_prctl.h" @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl_inval1(CPUArchState *env, abi_long arg2) #ifndef do_prctl_set_unalign #define do_prctl_set_unalign do_prctl_inval1 #endif +#ifndef do_prctl_sme_get_vl +#define do_prctl_sme_get_vl do_prctl_inval0 +#endif +#ifndef do_prctl_sme_set_vl +#define do_prctl_sme_set_vl do_prctl_inval1 +#endif static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, abi_long arg3, abi_long arg4, abi_long arg5) @@ -XXX,XX +XXX,XX @@ static abi_long do_prctl(CPUArchState *env, abi_long option, abi_long arg2, return do_prctl_sve_get_vl(env); case PR_SVE_SET_VL: return do_prctl_sve_set_vl(env, arg2); + case PR_SME_GET_VL: + return do_prctl_sme_get_vl(env); + case PR_SME_SET_VL: + return do_prctl_sme_set_vl(env, arg2); case PR_PAC_RESET_KEYS: if (arg3 || arg4 || arg5) { return -TARGET_EINVAL; -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> There's no reason to set CPACR_EL1.ZEN if SVE disabled. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-44-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(DeviceState *dev) /* and to the FP/Neon instructions */ env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, CPACR_EL1, FPEN, 3); - /* and to the SVE instructions */ - env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, - CPACR_EL1, ZEN, 3); - /* with reasonable vector length */ + /* and to the SVE instructions, with default vector length */ if (cpu_isar_feature(aa64_sve, cpu)) { + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } /* -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Enable SME, TPIDR2_EL0, and FA64 if supported by the cpu. Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-45-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(DeviceState *dev) CPACR_EL1, ZEN, 3); env->vfp.zcr_el[1] = cpu->sve_default_vq - 1; } + /* and for SME instructions, with default vector length, and TPIDR2 */ + if (cpu_isar_feature(aa64_sme, cpu)) { + env->cp15.sctlr_el[1] |= SCTLR_EnTP2; + env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, + CPACR_EL1, SMEN, 3); + env->vfp.smcr_el[1] = cpu->sme_default_vq - 1; + if (cpu_isar_feature(aa64_sme_fa64, cpu)) { + env->vfp.smcr_el[1] = FIELD_DP64(env->vfp.smcr_el[1], + SMCR, FA64, 1); + } + } /* * Enable 48-bit address space (TODO: take reserved_va into account). * Enable TBI0 but not TBI1. -- 2.25.1 From: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20220708151540.18136-46-richard.henderson@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- linux-user/elfload.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/linux-user/elfload.c b/linux-user/elfload.c index XXXXXXX..XXXXXXX 100644 --- a/linux-user/elfload.c +++ b/linux-user/elfload.c @@ -XXX,XX +XXX,XX @@ enum { ARM_HWCAP2_A64_RNG = 1 << 16, ARM_HWCAP2_A64_BTI = 1 << 17, ARM_HWCAP2_A64_MTE = 1 << 18, + ARM_HWCAP2_A64_ECV = 1 << 19, + ARM_HWCAP2_A64_AFP = 1 << 20, + ARM_HWCAP2_A64_RPRES = 1 << 21, + ARM_HWCAP2_A64_MTE3 = 1 << 22, + ARM_HWCAP2_A64_SME = 1 << 23, + ARM_HWCAP2_A64_SME_I16I64 = 1 << 24, + ARM_HWCAP2_A64_SME_F64F64 = 1 << 25, + ARM_HWCAP2_A64_SME_I8I32 = 1 << 26, + ARM_HWCAP2_A64_SME_F16F32 = 1 << 27, + ARM_HWCAP2_A64_SME_B16F32 = 1 << 28, + ARM_HWCAP2_A64_SME_F32F32 = 1 << 29, + ARM_HWCAP2_A64_SME_FA64 = 1 << 30, }; #define ELF_HWCAP get_elf_hwcap() @@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap2(void) GET_FEATURE_ID(aa64_rndr, ARM_HWCAP2_A64_RNG); GET_FEATURE_ID(aa64_bti, ARM_HWCAP2_A64_BTI); GET_FEATURE_ID(aa64_mte, ARM_HWCAP2_A64_MTE); + GET_FEATURE_ID(aa64_sme, (ARM_HWCAP2_A64_SME | + ARM_HWCAP2_A64_SME_F32F32 | + ARM_HWCAP2_A64_SME_B16F32 | + ARM_HWCAP2_A64_SME_F16F32 | + ARM_HWCAP2_A64_SME_I8I32)); + GET_FEATURE_ID(aa64_sme_f64f64, ARM_HWCAP2_A64_SME_F64F64); + GET_FEATURE_ID(aa64_sme_i16i64, ARM_HWCAP2_A64_SME_I16I64); + GET_FEATURE_ID(aa64_sme_fa64, ARM_HWCAP2_A64_SME_FA64); return hwcaps; } -- 2.25.1 Some arm patches; my to-review queue is by no means empty, but this is a big enough set of patches to be getting on with... -- PMM The following changes since commit cb9c6a8e5ad6a1f0ce164d352e3102df46986e22: .gitlab-ci.d/windows: Work-around timeout and OpenGL problems of the MSYS2 jobs (2023-01-04 18:58:33 +0000) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20230105 for you to fetch changes up to 93c9678de9dc7d2e68f9e8477da072bac30ef132: hw/net: Fix read of uninitialized memory in imx_fec. (2023-01-05 15:33:00 +0000) ---------------------------------------------------------------- target-arm queue: * Implement AArch32 ARMv8-R support * Add Cortex-R52 CPU * fix handling of HLT semihosting in system mode * hw/timer/ixm_epit: cleanup and fix bug in compare handling * target/arm: Coding style fixes * target/arm: Clean up includes * nseries: minor code cleanups * target/arm: align exposed ID registers with Linux * hw/arm/smmu-common: remove unnecessary inlines * i.MX7D: Handle GPT timers * i.MX7D: Connect IRQs to GPIO devices * i.MX6UL: Add a specific GPT timer instance * hw/net: Fix read of uninitialized memory in imx_fec ---------------------------------------------------------------- Alex Bennée (1): target/arm: fix handling of HLT semihosting in system mode Axel Heider (8): hw/timer/imx_epit: improve comments hw/timer/imx_epit: cleanup CR defines hw/timer/imx_epit: define SR_OCIF hw/timer/imx_epit: update interrupt state on CR write access hw/timer/imx_epit: hard reset initializes CR with 0 hw/timer/imx_epit: factor out register write handlers hw/timer/imx_epit: remove explicit fields cnt and freq hw/timer/imx_epit: fix compare timer handling Claudio Fontana (1): target/arm: cleanup cpu includes Fabiano Rosas (5): target/arm: Fix checkpatch comment style warnings in helper.c target/arm: Fix checkpatch space errors in helper.c target/arm: Fix checkpatch brace errors in helper.c target/arm: Remove unused includes from m_helper.c target/arm: Remove unused includes from helper.c Jean-Christophe Dubois (4): i.MX7D: Connect GPT timers to IRQ i.MX7D: Compute clock frequency for the fixed frequency clocks. i.MX6UL: Add a specific GPT timer instance for the i.MX6UL i.MX7D: Connect IRQs to GPIO devices. Peter Maydell (1): target/arm:Set lg_page_size to 0 if either S1 or S2 asks for it Philippe Mathieu-Daudé (5): hw/input/tsc2xxx: Constify set_transform()'s MouseTransformInfo arg hw/arm/nseries: Constify various read-only arrays hw/arm/nseries: Silent -Wmissing-field-initializers warning hw/arm/smmu-common: Reduce smmu_inv_notifiers_mr() scope hw/arm/smmu-common: Avoid using inlined functions with external linkage Stephen Longfield (1): hw/net: Fix read of uninitialized memory in imx_fec. Tobias Röhmel (7): target/arm: Don't add all MIDR aliases for cores that implement PMSA target/arm: Make RVBAR available for all ARMv8 CPUs target/arm: Make stage_2_format for cache attributes optional target/arm: Enable TTBCR_EAE for ARMv8-R AArch32 target/arm: Add PMSAv8r registers target/arm: Add PMSAv8r functionality target/arm: Add ARM Cortex-R52 CPU Zhuojia Shen (1): target/arm: align exposed ID registers with Linux include/hw/arm/fsl-imx7.h | 20 + include/hw/arm/smmu-common.h | 3 - include/hw/input/tsc2xxx.h | 4 +- include/hw/timer/imx_epit.h | 8 +- include/hw/timer/imx_gpt.h | 1 + target/arm/cpu.h | 6 + target/arm/internals.h | 4 + hw/arm/fsl-imx6ul.c | 2 +- hw/arm/fsl-imx7.c | 41 +- hw/arm/nseries.c | 28 +- hw/arm/smmu-common.c | 15 +- hw/input/tsc2005.c | 2 +- hw/input/tsc210x.c | 3 +- hw/misc/imx6ul_ccm.c | 6 - hw/misc/imx7_ccm.c | 49 ++- hw/net/imx_fec.c | 8 +- hw/timer/imx_epit.c | 376 +++++++++------- hw/timer/imx_gpt.c | 25 ++ target/arm/cpu.c | 35 +- target/arm/cpu64.c | 6 - target/arm/cpu_tcg.c | 42 ++ target/arm/debug_helper.c | 3 + target/arm/helper.c | 871 +++++++++++++++++++++++++++++--------- target/arm/m_helper.c | 16 - target/arm/machine.c | 28 ++ target/arm/ptw.c | 152 +++++-- target/arm/tlb_helper.c | 4 + target/arm/translate.c | 2 +- tests/tcg/aarch64/sysregs.c | 24 +- tests/tcg/aarch64/Makefile.target | 7 +- 30 files changed, 1330 insertions(+), 461 deletions(-) In get_phys_addr_twostage() we set the lg_page_size of the result to the maximum of the stage 1 and stage 2 page sizes. This works for the case where we do want to create a TLB entry, because we know the common TLB code only creates entries of the TARGET_PAGE_SIZE and asking for a size larger than that only means that invalidations invalidate the whole larger area. However, if lg_page_size is smaller than TARGET_PAGE_SIZE this effectively means "don't create a TLB entry"; in this case if either S1 or S2 said "this covers less than a page and can't go in a TLB" then the final result also should be marked that way. Set the resulting page size to 0 if either stage asked for a less-than-a-page entry, and expand the comment to explain what's going on. This has no effect for VMSA because currently the VMSA lookup always returns results that cover at least TARGET_PAGE_SIZE; however when we add v8R support it will reuse this code path, and for v8R the S1 and S2 results can be smaller than TARGET_PAGE_SIZE. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20221212142708.610090-1-peter.maydell@linaro.org --- target/arm/ptw.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/target/arm/ptw.c b/target/arm/ptw.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/ptw.c +++ b/target/arm/ptw.c @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_twostage(CPUARMState *env, S1Translate *ptw, } /* - * Use the maximum of the S1 & S2 page size, so that invalidation - * of pages > TARGET_PAGE_SIZE works correctly. + * If either S1 or S2 returned a result smaller than TARGET_PAGE_SIZE, + * this means "don't put this in the TLB"; in this case, return a + * result with lg_page_size == 0 to achieve that. Otherwise, + * use the maximum of the S1 & S2 page size, so that invalidation + * of pages > TARGET_PAGE_SIZE works correctly. (This works even though + * we know the combined result permissions etc only cover the minimum + * of the S1 and S2 page size, because we know that the common TLB code + * never actually creates TLB entries bigger than TARGET_PAGE_SIZE, + * and passing a larger page size value only affects invalidations.) */ - if (result->f.lg_page_size < s1_lgpgsz) { + if (result->f.lg_page_size < TARGET_PAGE_BITS || + s1_lgpgsz < TARGET_PAGE_BITS) { + result->f.lg_page_size = 0; + } else if (result->f.lg_page_size < s1_lgpgsz) { result->f.lg_page_size = s1_lgpgsz; } -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Cores with PMSA have the MPUIR register which has the same encoding as the MIDR alias with opc2=4. So we only add that alias if we are not realizing a core that implements PMSA. Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20221206102504.165775-2-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .access = PL1_R, .type = ARM_CP_NO_RAW, .resetvalue = cpu->midr, .fieldoffset = offsetof(CPUARMState, cp15.c0_cpuid), .readfn = midr_read }, - /* crn = 0 op1 = 0 crm = 0 op2 = 4,7 : AArch32 aliases of MIDR */ - { .name = "MIDR", .type = ARM_CP_ALIAS | ARM_CP_CONST, - .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 4, - .access = PL1_R, .resetvalue = cpu->midr }, + /* crn = 0 op1 = 0 crm = 0 op2 = 7 : AArch32 aliases of MIDR */ { .name = "MIDR", .type = ARM_CP_ALIAS | ARM_CP_CONST, .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 7, .access = PL1_R, .resetvalue = cpu->midr }, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .accessfn = access_aa64_tid1, .type = ARM_CP_CONST, .resetvalue = cpu->revidr }, }; + ARMCPRegInfo id_v8_midr_alias_cp_reginfo = { + .name = "MIDR", .type = ARM_CP_ALIAS | ARM_CP_CONST, + .cp = 15, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 4, + .access = PL1_R, .resetvalue = cpu->midr + }; ARMCPRegInfo id_cp_reginfo[] = { /* These are common to v8 and pre-v8 */ { .name = "CTR", @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) } if (arm_feature(env, ARM_FEATURE_V8)) { define_arm_cp_regs(cpu, id_v8_midr_cp_reginfo); + if (!arm_feature(env, ARM_FEATURE_PMSA)) { + define_one_arm_cp_reg(cpu, &id_v8_midr_alias_cp_reginfo); + } } else { define_arm_cp_regs(cpu, id_pre_v8_midr_cp_reginfo); } -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> RVBAR shadows RVBAR_ELx where x is the highest exception level if the highest EL is not EL3. This patch also allows ARMv8 CPUs to change the reset address with the rvbar property. Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20221206102504.165775-3-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 6 +++++- target/arm/helper.c | 21 ++++++++++++++------- 2 files changed, 19 insertions(+), 8 deletions(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset_hold(Object *obj) env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1, CPACR, CP11, 3); #endif + if (arm_feature(env, ARM_FEATURE_V8)) { + env->cp15.rvbar = cpu->rvbar_prop; + env->regs[15] = cpu->rvbar_prop; + } } #if defined(CONFIG_USER_ONLY) @@ -XXX,XX +XXX,XX @@ void arm_cpu_post_init(Object *obj) qdev_property_add_static(DEVICE(obj), &arm_cpu_reset_hivecs_property); } - if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64)) { + if (arm_feature(&cpu->env, ARM_FEATURE_V8)) { object_property_add_uint64_ptr(obj, "rvbar", &cpu->rvbar_prop, OBJ_PROP_FLAG_READWRITE); diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) if (!arm_feature(env, ARM_FEATURE_EL3) && !arm_feature(env, ARM_FEATURE_EL2)) { ARMCPRegInfo rvbar = { - .name = "RVBAR_EL1", .state = ARM_CP_STATE_AA64, + .name = "RVBAR_EL1", .state = ARM_CP_STATE_BOTH, .opc0 = 3, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1, .access = PL1_R, .fieldoffset = offsetof(CPUARMState, cp15.rvbar), @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) } /* RVBAR_EL2 is only implemented if EL2 is the highest EL */ if (!arm_feature(env, ARM_FEATURE_EL3)) { - ARMCPRegInfo rvbar = { - .name = "RVBAR_EL2", .state = ARM_CP_STATE_AA64, - .opc0 = 3, .opc1 = 4, .crn = 12, .crm = 0, .opc2 = 1, - .access = PL2_R, - .fieldoffset = offsetof(CPUARMState, cp15.rvbar), + ARMCPRegInfo rvbar[] = { + { + .name = "RVBAR_EL2", .state = ARM_CP_STATE_AA64, + .opc0 = 3, .opc1 = 4, .crn = 12, .crm = 0, .opc2 = 1, + .access = PL2_R, + .fieldoffset = offsetof(CPUARMState, cp15.rvbar), + }, + { .name = "RVBAR", .type = ARM_CP_ALIAS, + .cp = 15, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1, + .access = PL2_R, + .fieldoffset = offsetof(CPUARMState, cp15.rvbar), + }, }; - define_one_arm_cp_reg(cpu, &rvbar); + define_arm_cp_regs(cpu, rvbar); } } -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> The v8R PMSAv8 has a two-stage MPU translation process, but, unlike VMSAv8, the stage 2 attributes are in the same format as the stage 1 attributes (8-bit MAIR format). Rather than converting the MAIR format to the format used for VMSA stage 2 (bits [5:2] of a VMSA stage 2 descriptor) and then converting back to do the attribute combination, allow combined_attrs_nofwb() to accept s2 attributes that are already in the MAIR format. We move the assert() to combined_attrs_fwb(), because that function really does require a VMSA stage 2 attribute format. (We will never get there for v8R, because PMSAv8 does not implement FEAT_S2FWB.) Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20221206102504.165775-4-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/ptw.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/target/arm/ptw.c b/target/arm/ptw.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/ptw.c +++ b/target/arm/ptw.c @@ -XXX,XX +XXX,XX @@ static uint8_t combined_attrs_nofwb(uint64_t hcr, { uint8_t s1lo, s2lo, s1hi, s2hi, s2_mair_attrs, ret_attrs; - s2_mair_attrs = convert_stage2_attrs(hcr, s2.attrs); + if (s2.is_s2_format) { + s2_mair_attrs = convert_stage2_attrs(hcr, s2.attrs); + } else { + s2_mair_attrs = s2.attrs; + } s1lo = extract32(s1.attrs, 0, 4); s2lo = extract32(s2_mair_attrs, 0, 4); @@ -XXX,XX +XXX,XX @@ static uint8_t force_cacheattr_nibble_wb(uint8_t attr) */ static uint8_t combined_attrs_fwb(ARMCacheAttrs s1, ARMCacheAttrs s2) { + assert(s2.is_s2_format && !s1.is_s2_format); + switch (s2.attrs) { case 7: /* Use stage 1 attributes */ @@ -XXX,XX +XXX,XX @@ static ARMCacheAttrs combine_cacheattrs(uint64_t hcr, ARMCacheAttrs ret; bool tagged = false; - assert(s2.is_s2_format && !s1.is_s2_format); + assert(!s1.is_s2_format); ret.is_s2_format = false; if (s1.attrs == 0xf0) { -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> ARMv8-R AArch32 CPUs behave as if TTBCR.EAE is always 1 even tough they don't have the TTBCR register. See ARM Architecture Reference Manual Supplement - ARMv8, for the ARMv8-R AArch32 architecture profile Version:A.c section C1.2. Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20221206102504.165775-5-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/internals.h | 4 ++++ target/arm/debug_helper.c | 3 +++ target/arm/tlb_helper.c | 4 ++++ 3 files changed, 11 insertions(+) diff --git a/target/arm/internals.h b/target/arm/internals.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/internals.h +++ b/target/arm/internals.h @@ -XXX,XX +XXX,XX @@ unsigned int arm_pamax(ARMCPU *cpu); static inline bool extended_addresses_enabled(CPUARMState *env) { uint64_t tcr = env->cp15.tcr_el[arm_is_secure(env) ? 3 : 1]; + if (arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8)) { + return true; + } return arm_el_is_aa64(env, 1) || (arm_feature(env, ARM_FEATURE_LPAE) && (tcr & TTBCR_EAE)); } diff --git a/target/arm/debug_helper.c b/target/arm/debug_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/debug_helper.c +++ b/target/arm/debug_helper.c @@ -XXX,XX +XXX,XX @@ static uint32_t arm_debug_exception_fsr(CPUARMState *env) if (target_el == 2 || arm_el_is_aa64(env, target_el)) { using_lpae = true; + } else if (arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8)) { + using_lpae = true; } else { if (arm_feature(env, ARM_FEATURE_LPAE) && (env->cp15.tcr_el[target_el] & TTBCR_EAE)) { diff --git a/target/arm/tlb_helper.c b/target/arm/tlb_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tlb_helper.c +++ b/target/arm/tlb_helper.c @@ -XXX,XX +XXX,XX @@ bool regime_using_lpae_format(CPUARMState *env, ARMMMUIdx mmu_idx) if (el == 2 || arm_el_is_aa64(env, el)) { return true; } + if (arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8)) { + return true; + } if (arm_feature(env, ARM_FEATURE_LPAE) && (regime_tcr(env, mmu_idx) & TTBCR_EAE)) { return true; -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Message-id: 20221206102504.165775-6-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 6 + target/arm/cpu.c | 28 +++- target/arm/helper.c | 302 +++++++++++++++++++++++++++++++++++++++++++ target/arm/machine.c | 28 ++++ 4 files changed, 360 insertions(+), 4 deletions(-) diff --git a/target/arm/cpu.h b/target/arm/cpu.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.h +++ b/target/arm/cpu.h @@ -XXX,XX +XXX,XX @@ typedef struct CPUArchState { }; uint64_t sctlr_el[4]; }; + uint64_t vsctlr; /* Virtualization System control register. */ uint64_t cpacr_el1; /* Architectural feature access control register */ uint64_t cptr_el[4]; /* ARMv8 feature trap registers */ uint32_t c1_xscaleauxcr; /* XScale auxiliary control register. */ @@ -XXX,XX +XXX,XX @@ typedef struct CPUArchState { */ uint32_t *rbar[M_REG_NUM_BANKS]; uint32_t *rlar[M_REG_NUM_BANKS]; + uint32_t *hprbar; + uint32_t *hprlar; uint32_t mair0[M_REG_NUM_BANKS]; uint32_t mair1[M_REG_NUM_BANKS]; + uint32_t hprselr; } pmsav8; /* v8M SAU */ @@ -XXX,XX +XXX,XX @@ struct ArchCPU { bool has_mpu; /* PMSAv7 MPU number of supported regions */ uint32_t pmsav7_dregion; + /* PMSAv8 MPU number of supported hyp regions */ + uint32_t pmsav8r_hdregion; /* v8M SAU number of supported regions */ uint32_t sau_sregion; diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset_hold(Object *obj) sizeof(*env->pmsav7.dracr) * cpu->pmsav7_dregion); } } + + if (cpu->pmsav8r_hdregion > 0) { + memset(env->pmsav8.hprbar, 0, + sizeof(*env->pmsav8.hprbar) * cpu->pmsav8r_hdregion); + memset(env->pmsav8.hprlar, 0, + sizeof(*env->pmsav8.hprlar) * cpu->pmsav8r_hdregion); + } + env->pmsav7.rnr[M_REG_NS] = 0; env->pmsav7.rnr[M_REG_S] = 0; env->pmsav8.mair0[M_REG_NS] = 0; @@ -XXX,XX +XXX,XX @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp) /* MPU can be configured out of a PMSA CPU either by setting has-mpu * to false or by setting pmsav7-dregion to 0. */ - if (!cpu->has_mpu) { - cpu->pmsav7_dregion = 0; - } - if (cpu->pmsav7_dregion == 0) { + if (!cpu->has_mpu || cpu->pmsav7_dregion == 0) { cpu->has_mpu = false; + cpu->pmsav7_dregion = 0; + cpu->pmsav8r_hdregion = 0; } if (arm_feature(env, ARM_FEATURE_PMSA) && @@ -XXX,XX +XXX,XX @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp) env->pmsav7.dracr = g_new0(uint32_t, nr); } } + + if (cpu->pmsav8r_hdregion > 0xff) { + error_setg(errp, "PMSAv8 MPU EL2 #regions invalid %" PRIu32, + cpu->pmsav8r_hdregion); + return; + } + + if (cpu->pmsav8r_hdregion) { + env->pmsav8.hprbar = g_new0(uint32_t, + cpu->pmsav8r_hdregion); + env->pmsav8.hprlar = g_new0(uint32_t, + cpu->pmsav8r_hdregion); + } } if (arm_feature(env, ARM_FEATURE_M_SECURITY)) { diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ static void pmsav7_rgnr_write(CPUARMState *env, const ARMCPRegInfo *ri, raw_write(env, ri, value); } +static void prbar_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + env->pmsav8.rbar[M_REG_NS][env->pmsav7.rnr[M_REG_NS]] = value; +} + +static uint64_t prbar_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + return env->pmsav8.rbar[M_REG_NS][env->pmsav7.rnr[M_REG_NS]]; +} + +static void prlar_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + env->pmsav8.rlar[M_REG_NS][env->pmsav7.rnr[M_REG_NS]] = value; +} + +static uint64_t prlar_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + return env->pmsav8.rlar[M_REG_NS][env->pmsav7.rnr[M_REG_NS]]; +} + +static void prselr_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + /* + * Ignore writes that would select not implemented region. + * This is architecturally UNPREDICTABLE. + */ + if (value >= cpu->pmsav7_dregion) { + return; + } + + env->pmsav7.rnr[M_REG_NS] = value; +} + +static void hprbar_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + env->pmsav8.hprbar[env->pmsav8.hprselr] = value; +} + +static uint64_t hprbar_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + return env->pmsav8.hprbar[env->pmsav8.hprselr]; +} + +static void hprlar_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + env->pmsav8.hprlar[env->pmsav8.hprselr] = value; +} + +static uint64_t hprlar_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + return env->pmsav8.hprlar[env->pmsav8.hprselr]; +} + +static void hprenr_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + uint32_t n; + uint32_t bit; + ARMCPU *cpu = env_archcpu(env); + + /* Ignore writes to unimplemented regions */ + int rmax = MIN(cpu->pmsav8r_hdregion, 32); + value &= MAKE_64BIT_MASK(0, rmax); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + + /* Register alias is only valid for first 32 indexes */ + for (n = 0; n < rmax; ++n) { + bit = extract32(value, n, 1); + env->pmsav8.hprlar[n] = deposit32( + env->pmsav8.hprlar[n], 0, 1, bit); + } +} + +static uint64_t hprenr_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + uint32_t n; + uint32_t result = 0x0; + ARMCPU *cpu = env_archcpu(env); + + /* Register alias is only valid for first 32 indexes */ + for (n = 0; n < MIN(cpu->pmsav8r_hdregion, 32); ++n) { + if (env->pmsav8.hprlar[n] & 0x1) { + result |= (0x1 << n); + } + } + return result; +} + +static void hprselr_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + + /* + * Ignore writes that would select not implemented region. + * This is architecturally UNPREDICTABLE. + */ + if (value >= cpu->pmsav8r_hdregion) { + return; + } + + env->pmsav8.hprselr = value; +} + +static void pmsav8r_regn_write(CPUARMState *env, const ARMCPRegInfo *ri, + uint64_t value) +{ + ARMCPU *cpu = env_archcpu(env); + uint8_t index = (extract32(ri->opc0, 0, 1) << 4) | + (extract32(ri->crm, 0, 3) << 1) | extract32(ri->opc2, 2, 1); + + tlb_flush(CPU(cpu)); /* Mappings may have changed - purge! */ + + if (ri->opc1 & 4) { + if (index >= cpu->pmsav8r_hdregion) { + return; + } + if (ri->opc2 & 0x1) { + env->pmsav8.hprlar[index] = value; + } else { + env->pmsav8.hprbar[index] = value; + } + } else { + if (index >= cpu->pmsav7_dregion) { + return; + } + if (ri->opc2 & 0x1) { + env->pmsav8.rlar[M_REG_NS][index] = value; + } else { + env->pmsav8.rbar[M_REG_NS][index] = value; + } + } +} + +static uint64_t pmsav8r_regn_read(CPUARMState *env, const ARMCPRegInfo *ri) +{ + ARMCPU *cpu = env_archcpu(env); + uint8_t index = (extract32(ri->opc0, 0, 1) << 4) | + (extract32(ri->crm, 0, 3) << 1) | extract32(ri->opc2, 2, 1); + + if (ri->opc1 & 4) { + if (index >= cpu->pmsav8r_hdregion) { + return 0x0; + } + if (ri->opc2 & 0x1) { + return env->pmsav8.hprlar[index]; + } else { + return env->pmsav8.hprbar[index]; + } + } else { + if (index >= cpu->pmsav7_dregion) { + return 0x0; + } + if (ri->opc2 & 0x1) { + return env->pmsav8.rlar[M_REG_NS][index]; + } else { + return env->pmsav8.rbar[M_REG_NS][index]; + } + } +} + +static const ARMCPRegInfo pmsav8r_cp_reginfo[] = { + { .name = "PRBAR", + .cp = 15, .opc1 = 0, .crn = 6, .crm = 3, .opc2 = 0, + .access = PL1_RW, .type = ARM_CP_NO_RAW, + .accessfn = access_tvm_trvm, + .readfn = prbar_read, .writefn = prbar_write }, + { .name = "PRLAR", + .cp = 15, .opc1 = 0, .crn = 6, .crm = 3, .opc2 = 1, + .access = PL1_RW, .type = ARM_CP_NO_RAW, + .accessfn = access_tvm_trvm, + .readfn = prlar_read, .writefn = prlar_write }, + { .name = "PRSELR", .resetvalue = 0, + .cp = 15, .opc1 = 0, .crn = 6, .crm = 2, .opc2 = 1, + .access = PL1_RW, .accessfn = access_tvm_trvm, + .writefn = prselr_write, + .fieldoffset = offsetof(CPUARMState, pmsav7.rnr[M_REG_NS]) }, + { .name = "HPRBAR", .resetvalue = 0, + .cp = 15, .opc1 = 4, .crn = 6, .crm = 3, .opc2 = 0, + .access = PL2_RW, .type = ARM_CP_NO_RAW, + .readfn = hprbar_read, .writefn = hprbar_write }, + { .name = "HPRLAR", + .cp = 15, .opc1 = 4, .crn = 6, .crm = 3, .opc2 = 1, + .access = PL2_RW, .type = ARM_CP_NO_RAW, + .readfn = hprlar_read, .writefn = hprlar_write }, + { .name = "HPRSELR", .resetvalue = 0, + .cp = 15, .opc1 = 4, .crn = 6, .crm = 2, .opc2 = 1, + .access = PL2_RW, + .writefn = hprselr_write, + .fieldoffset = offsetof(CPUARMState, pmsav8.hprselr) }, + { .name = "HPRENR", + .cp = 15, .opc1 = 4, .crn = 6, .crm = 1, .opc2 = 1, + .access = PL2_RW, .type = ARM_CP_NO_RAW, + .readfn = hprenr_read, .writefn = hprenr_write }, +}; + static const ARMCPRegInfo pmsav7_cp_reginfo[] = { /* Reset for all these registers is handled in arm_cpu_reset(), * because the PMSAv7 is also used by M-profile CPUs, which do @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .access = PL1_R, .type = ARM_CP_CONST, .resetvalue = cpu->pmsav7_dregion << 8 }; + /* HMPUIR is specific to PMSA V8 */ + ARMCPRegInfo id_hmpuir_reginfo = { + .name = "HMPUIR", + .cp = 15, .opc1 = 4, .crn = 0, .crm = 0, .opc2 = 4, + .access = PL2_R, .type = ARM_CP_CONST, + .resetvalue = cpu->pmsav8r_hdregion + }; static const ARMCPRegInfo crn0_wi_reginfo = { .name = "CRN0_WI", .cp = 15, .crn = 0, .crm = CP_ANY, .opc1 = CP_ANY, .opc2 = CP_ANY, .access = PL1_W, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) define_arm_cp_regs(cpu, id_cp_reginfo); if (!arm_feature(env, ARM_FEATURE_PMSA)) { define_one_arm_cp_reg(cpu, &id_tlbtr_reginfo); + } else if (arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8)) { + uint32_t i = 0; + char *tmp_string; + + define_one_arm_cp_reg(cpu, &id_mpuir_reginfo); + define_one_arm_cp_reg(cpu, &id_hmpuir_reginfo); + define_arm_cp_regs(cpu, pmsav8r_cp_reginfo); + + /* Register alias is only valid for first 32 indexes */ + for (i = 0; i < MIN(cpu->pmsav7_dregion, 32); ++i) { + uint8_t crm = 0b1000 | extract32(i, 1, 3); + uint8_t opc1 = extract32(i, 4, 1); + uint8_t opc2 = extract32(i, 0, 1) << 2; + + tmp_string = g_strdup_printf("PRBAR%u", i); + ARMCPRegInfo tmp_prbarn_reginfo = { + .name = tmp_string, .type = ARM_CP_ALIAS | ARM_CP_NO_RAW, + .cp = 15, .opc1 = opc1, .crn = 6, .crm = crm, .opc2 = opc2, + .access = PL1_RW, .resetvalue = 0, + .accessfn = access_tvm_trvm, + .writefn = pmsav8r_regn_write, .readfn = pmsav8r_regn_read + }; + define_one_arm_cp_reg(cpu, &tmp_prbarn_reginfo); + g_free(tmp_string); + + opc2 = extract32(i, 0, 1) << 2 | 0x1; + tmp_string = g_strdup_printf("PRLAR%u", i); + ARMCPRegInfo tmp_prlarn_reginfo = { + .name = tmp_string, .type = ARM_CP_ALIAS | ARM_CP_NO_RAW, + .cp = 15, .opc1 = opc1, .crn = 6, .crm = crm, .opc2 = opc2, + .access = PL1_RW, .resetvalue = 0, + .accessfn = access_tvm_trvm, + .writefn = pmsav8r_regn_write, .readfn = pmsav8r_regn_read + }; + define_one_arm_cp_reg(cpu, &tmp_prlarn_reginfo); + g_free(tmp_string); + } + + /* Register alias is only valid for first 32 indexes */ + for (i = 0; i < MIN(cpu->pmsav8r_hdregion, 32); ++i) { + uint8_t crm = 0b1000 | extract32(i, 1, 3); + uint8_t opc1 = 0b100 | extract32(i, 4, 1); + uint8_t opc2 = extract32(i, 0, 1) << 2; + + tmp_string = g_strdup_printf("HPRBAR%u", i); + ARMCPRegInfo tmp_hprbarn_reginfo = { + .name = tmp_string, + .type = ARM_CP_NO_RAW, + .cp = 15, .opc1 = opc1, .crn = 6, .crm = crm, .opc2 = opc2, + .access = PL2_RW, .resetvalue = 0, + .writefn = pmsav8r_regn_write, .readfn = pmsav8r_regn_read + }; + define_one_arm_cp_reg(cpu, &tmp_hprbarn_reginfo); + g_free(tmp_string); + + opc2 = extract32(i, 0, 1) << 2 | 0x1; + tmp_string = g_strdup_printf("HPRLAR%u", i); + ARMCPRegInfo tmp_hprlarn_reginfo = { + .name = tmp_string, + .type = ARM_CP_NO_RAW, + .cp = 15, .opc1 = opc1, .crn = 6, .crm = crm, .opc2 = opc2, + .access = PL2_RW, .resetvalue = 0, + .writefn = pmsav8r_regn_write, .readfn = pmsav8r_regn_read + }; + define_one_arm_cp_reg(cpu, &tmp_hprlarn_reginfo); + g_free(tmp_string); + } } else if (arm_feature(env, ARM_FEATURE_V7)) { define_one_arm_cp_reg(cpu, &id_mpuir_reginfo); } @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) sctlr.type |= ARM_CP_SUPPRESS_TB_END; } define_one_arm_cp_reg(cpu, &sctlr); + + if (arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8)) { + ARMCPRegInfo vsctlr = { + .name = "VSCTLR", .state = ARM_CP_STATE_AA32, + .cp = 15, .opc1 = 4, .crn = 2, .crm = 0, .opc2 = 0, + .access = PL2_RW, .resetvalue = 0x0, + .fieldoffset = offsetoflow32(CPUARMState, cp15.vsctlr), + }; + define_one_arm_cp_reg(cpu, &vsctlr); + } } if (cpu_isar_feature(aa64_lor, cpu)) { diff --git a/target/arm/machine.c b/target/arm/machine.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/machine.c +++ b/target/arm/machine.c @@ -XXX,XX +XXX,XX @@ static bool pmsav8_needed(void *opaque) arm_feature(env, ARM_FEATURE_V8); } +static bool pmsav8r_needed(void *opaque) +{ + ARMCPU *cpu = opaque; + CPUARMState *env = &cpu->env; + + return arm_feature(env, ARM_FEATURE_PMSA) && + arm_feature(env, ARM_FEATURE_V8) && + !arm_feature(env, ARM_FEATURE_M); +} + +static const VMStateDescription vmstate_pmsav8r = { + .name = "cpu/pmsav8/pmsav8r", + .version_id = 1, + .minimum_version_id = 1, + .needed = pmsav8r_needed, + .fields = (VMStateField[]) { + VMSTATE_VARRAY_UINT32(env.pmsav8.hprbar, ARMCPU, + pmsav8r_hdregion, 0, vmstate_info_uint32, uint32_t), + VMSTATE_VARRAY_UINT32(env.pmsav8.hprlar, ARMCPU, + pmsav8r_hdregion, 0, vmstate_info_uint32, uint32_t), + VMSTATE_END_OF_LIST() + }, +}; + static const VMStateDescription vmstate_pmsav8 = { .name = "cpu/pmsav8", .version_id = 1, @@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_pmsav8 = { VMSTATE_UINT32(env.pmsav8.mair0[M_REG_NS], ARMCPU), VMSTATE_UINT32(env.pmsav8.mair1[M_REG_NS], ARMCPU), VMSTATE_END_OF_LIST() + }, + .subsections = (const VMStateDescription * []) { + &vmstate_pmsav8r, + NULL } }; -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Add PMSAv8r translation. Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20221206102504.165775-7-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/ptw.c | 126 ++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 104 insertions(+), 22 deletions(-) diff --git a/target/arm/ptw.c b/target/arm/ptw.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/ptw.c +++ b/target/arm/ptw.c @@ -XXX,XX +XXX,XX @@ static bool pmsav7_use_background_region(ARMCPU *cpu, ARMMMUIdx mmu_idx, if (arm_feature(env, ARM_FEATURE_M)) { return env->v7m.mpu_ctrl[is_secure] & R_V7M_MPU_CTRL_PRIVDEFENA_MASK; - } else { - return regime_sctlr(env, mmu_idx) & SCTLR_BR; } + + if (mmu_idx == ARMMMUIdx_Stage2) { + return false; + } + + return regime_sctlr(env, mmu_idx) & SCTLR_BR; } static bool get_phys_addr_pmsav7(CPUARMState *env, uint32_t address, @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_pmsav7(CPUARMState *env, uint32_t address, return !(result->f.prot & (1 << access_type)); } +static uint32_t *regime_rbar(CPUARMState *env, ARMMMUIdx mmu_idx, + uint32_t secure) +{ + if (regime_el(env, mmu_idx) == 2) { + return env->pmsav8.hprbar; + } else { + return env->pmsav8.rbar[secure]; + } +} + +static uint32_t *regime_rlar(CPUARMState *env, ARMMMUIdx mmu_idx, + uint32_t secure) +{ + if (regime_el(env, mmu_idx) == 2) { + return env->pmsav8.hprlar; + } else { + return env->pmsav8.rlar[secure]; + } +} + bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, MMUAccessType access_type, ARMMMUIdx mmu_idx, bool secure, GetPhysAddrResult *result, @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, bool hit = false; uint32_t addr_page_base = address & TARGET_PAGE_MASK; uint32_t addr_page_limit = addr_page_base + (TARGET_PAGE_SIZE - 1); + int region_counter; + + if (regime_el(env, mmu_idx) == 2) { + region_counter = cpu->pmsav8r_hdregion; + } else { + region_counter = cpu->pmsav7_dregion; + } result->f.lg_page_size = TARGET_PAGE_BITS; result->f.phys_addr = address; @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, *mregion = -1; } + if (mmu_idx == ARMMMUIdx_Stage2) { + fi->stage2 = true; + } + /* * Unlike the ARM ARM pseudocode, we don't need to check whether this * was an exception vector read from the vector table (which is always @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, hit = true; } - for (n = (int)cpu->pmsav7_dregion - 1; n >= 0; n--) { + uint32_t bitmask; + if (arm_feature(env, ARM_FEATURE_M)) { + bitmask = 0x1f; + } else { + bitmask = 0x3f; + fi->level = 0; + } + + for (n = region_counter - 1; n >= 0; n--) { /* region search */ /* - * Note that the base address is bits [31:5] from the register - * with bits [4:0] all zeroes, but the limit address is bits - * [31:5] from the register with bits [4:0] all ones. + * Note that the base address is bits [31:x] from the register + * with bits [x-1:0] all zeroes, but the limit address is bits + * [31:x] from the register with bits [x:0] all ones. Where x is + * 5 for Cortex-M and 6 for Cortex-R */ - uint32_t base = env->pmsav8.rbar[secure][n] & ~0x1f; - uint32_t limit = env->pmsav8.rlar[secure][n] | 0x1f; + uint32_t base = regime_rbar(env, mmu_idx, secure)[n] & ~bitmask; + uint32_t limit = regime_rlar(env, mmu_idx, secure)[n] | bitmask; - if (!(env->pmsav8.rlar[secure][n] & 0x1)) { + if (!(regime_rlar(env, mmu_idx, secure)[n] & 0x1)) { /* Region disabled */ continue; } @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, * PMSAv7 where highest-numbered-region wins) */ fi->type = ARMFault_Permission; - fi->level = 1; + if (arm_feature(env, ARM_FEATURE_M)) { + fi->level = 1; + } return true; } @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, } if (!hit) { - /* background fault */ - fi->type = ARMFault_Background; + if (arm_feature(env, ARM_FEATURE_M)) { + fi->type = ARMFault_Background; + } else { + fi->type = ARMFault_Permission; + } return true; } @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, /* hit using the background region */ get_phys_addr_pmsav7_default(env, mmu_idx, address, &result->f.prot); } else { - uint32_t ap = extract32(env->pmsav8.rbar[secure][matchregion], 1, 2); - uint32_t xn = extract32(env->pmsav8.rbar[secure][matchregion], 0, 1); + uint32_t matched_rbar = regime_rbar(env, mmu_idx, secure)[matchregion]; + uint32_t matched_rlar = regime_rlar(env, mmu_idx, secure)[matchregion]; + uint32_t ap = extract32(matched_rbar, 1, 2); + uint32_t xn = extract32(matched_rbar, 0, 1); bool pxn = false; if (arm_feature(env, ARM_FEATURE_V8_1M)) { - pxn = extract32(env->pmsav8.rlar[secure][matchregion], 4, 1); + pxn = extract32(matched_rlar, 4, 1); } if (m_is_system_region(env, address)) { @@ -XXX,XX +XXX,XX @@ bool pmsav8_mpu_lookup(CPUARMState *env, uint32_t address, xn = 1; } - result->f.prot = simple_ap_to_rw_prot(env, mmu_idx, ap); + if (regime_el(env, mmu_idx) == 2) { + result->f.prot = simple_ap_to_rw_prot_is_user(ap, + mmu_idx != ARMMMUIdx_E2); + } else { + result->f.prot = simple_ap_to_rw_prot(env, mmu_idx, ap); + } + + if (!arm_feature(env, ARM_FEATURE_M)) { + uint8_t attrindx = extract32(matched_rlar, 1, 3); + uint64_t mair = env->cp15.mair_el[regime_el(env, mmu_idx)]; + uint8_t sh = extract32(matched_rlar, 3, 2); + + if (regime_sctlr(env, mmu_idx) & SCTLR_WXN && + result->f.prot & PAGE_WRITE && mmu_idx != ARMMMUIdx_Stage2) { + xn = 0x1; + } + + if ((regime_el(env, mmu_idx) == 1) && + regime_sctlr(env, mmu_idx) & SCTLR_UWXN && ap == 0x1) { + pxn = 0x1; + } + + result->cacheattrs.is_s2_format = false; + result->cacheattrs.attrs = extract64(mair, attrindx * 8, 8); + result->cacheattrs.shareability = sh; + } + if (result->f.prot && !xn && !(pxn && !is_user)) { result->f.prot |= PAGE_EXEC; } - /* - * We don't need to look the attribute up in the MAIR0/MAIR1 - * registers because that only tells us about cacheability. - */ + if (mregion) { *mregion = matchregion; } } fi->type = ARMFault_Permission; - fi->level = 1; + if (arm_feature(env, ARM_FEATURE_M)) { + fi->level = 1; + } return !(result->f.prot & (1 << access_type)); } @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_twostage(CPUARMState *env, S1Translate *ptw, cacheattrs1 = result->cacheattrs; memset(result, 0, sizeof(*result)); - ret = get_phys_addr_lpae(env, ptw, ipa, access_type, is_el0, result, fi); + if (arm_feature(env, ARM_FEATURE_PMSA)) { + ret = get_phys_addr_pmsav8(env, ipa, access_type, + ptw->in_mmu_idx, is_secure, result, fi); + } else { + ret = get_phys_addr_lpae(env, ptw, ipa, access_type, + is_el0, result, fi); + } fi->s2addr = ipa; /* Combine the S1 and S2 perms. */ -- 2.25.1 From: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> All constants are taken from the ARM Cortex-R52 Processor TRM Revision: r1p3 Signed-off-by: Tobias Röhmel <tobias.roehmel@rwth-aachen.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20221206102504.165775-8-tobias.roehmel@rwth-aachen.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu_tcg.c | 42 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) diff --git a/target/arm/cpu_tcg.c b/target/arm/cpu_tcg.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu_tcg.c +++ b/target/arm/cpu_tcg.c @@ -XXX,XX +XXX,XX @@ static void cortex_r5_initfn(Object *obj) define_arm_cp_regs(cpu, cortexr5_cp_reginfo); } +static void cortex_r52_initfn(Object *obj) +{ + ARMCPU *cpu = ARM_CPU(obj); + + set_feature(&cpu->env, ARM_FEATURE_V8); + set_feature(&cpu->env, ARM_FEATURE_EL2); + set_feature(&cpu->env, ARM_FEATURE_PMSA); + set_feature(&cpu->env, ARM_FEATURE_NEON); + set_feature(&cpu->env, ARM_FEATURE_GENERIC_TIMER); + cpu->midr = 0x411fd133; /* r1p3 */ + cpu->revidr = 0x00000000; + cpu->reset_fpsid = 0x41034023; + cpu->isar.mvfr0 = 0x10110222; + cpu->isar.mvfr1 = 0x12111111; + cpu->isar.mvfr2 = 0x00000043; + cpu->ctr = 0x8144c004; + cpu->reset_sctlr = 0x30c50838; + cpu->isar.id_pfr0 = 0x00000131; + cpu->isar.id_pfr1 = 0x10111001; + cpu->isar.id_dfr0 = 0x03010006; + cpu->id_afr0 = 0x00000000; + cpu->isar.id_mmfr0 = 0x00211040; + cpu->isar.id_mmfr1 = 0x40000000; + cpu->isar.id_mmfr2 = 0x01200000; + cpu->isar.id_mmfr3 = 0xf0102211; + cpu->isar.id_mmfr4 = 0x00000010; + cpu->isar.id_isar0 = 0x02101110; + cpu->isar.id_isar1 = 0x13112111; + cpu->isar.id_isar2 = 0x21232142; + cpu->isar.id_isar3 = 0x01112131; + cpu->isar.id_isar4 = 0x00010142; + cpu->isar.id_isar5 = 0x00010001; + cpu->isar.dbgdidr = 0x77168000; + cpu->clidr = (1 << 27) | (1 << 24) | 0x3; + cpu->ccsidr[0] = 0x700fe01a; /* 32KB L1 dcache */ + cpu->ccsidr[1] = 0x201fe00a; /* 32KB L1 icache */ + + cpu->pmsav7_dregion = 16; + cpu->pmsav8r_hdregion = 16; +} + static void cortex_r5f_initfn(Object *obj) { ARMCPU *cpu = ARM_CPU(obj); @@ -XXX,XX +XXX,XX @@ static const ARMCPUInfo arm_tcg_cpus[] = { .class_init = arm_v7m_class_init }, { .name = "cortex-r5", .initfn = cortex_r5_initfn }, { .name = "cortex-r5f", .initfn = cortex_r5f_initfn }, + { .name = "cortex-r52", .initfn = cortex_r52_initfn }, { .name = "ti925t", .initfn = ti925t_initfn }, { .name = "sa1100", .initfn = sa1100_initfn }, { .name = "sa1110", .initfn = sa1110_initfn }, -- 2.25.1 From: Alex Bennée <alex.bennee@linaro.org> The check semihosting_enabled() wants to know if the guest is currently in user mode. Unlike the other cases the test was inverted causing us to block semihosting calls in non-EL0 modes. Cc: qemu-stable@nongnu.org Fixes: 19b26317e9 (target/arm: Honour -semihosting-config userspace=on) Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/translate.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/target/arm/translate.c b/target/arm/translate.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate.c +++ b/target/arm/translate.c @@ -XXX,XX +XXX,XX @@ static inline void gen_hlt(DisasContext *s, int imm) * semihosting, to provide some semblance of security * (and for consistency with our 32-bit semihosting). */ - if (semihosting_enabled(s->current_el != 0) && + if (semihosting_enabled(s->current_el == 0) && (imm == (s->thumb ? 0x3c : 0xf000))) { gen_exception_internal_insn(s, EXCP_SEMIHOST); return; -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> Fix typos, add background information Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/imx_epit.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-) diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_set_freq(IMXEPITState *s) } } +/* + * This is called both on hardware (device) reset and software reset. + */ static void imx_epit_reset(DeviceState *dev) { IMXEPITState *s = IMX_EPIT(dev); - /* - * Soft reset doesn't touch some bits; hard reset clears them - */ + /* Soft reset doesn't touch some bits; hard reset clears them */ s->cr &= (CR_EN|CR_ENMOD|CR_STOPEN|CR_DOZEN|CR_WAITEN|CR_DBGEN); s->sr = 0; s->lr = EPIT_TIMER_MAX; @@ -XXX,XX +XXX,XX @@ static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, ptimer_transaction_begin(s->timer_cmp); ptimer_transaction_begin(s->timer_reload); + /* Update the frequency. Has been done already in case of a reset. */ if (!(s->cr & CR_SWR)) { imx_epit_set_freq(s); } @@ -XXX,XX +XXX,XX @@ static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, break; case 1: /* SR - ACK*/ - /* writing 1 to OCIF clear the OCIF bit */ + /* writing 1 to OCIF clears the OCIF bit */ if (value & 0x01) { s->sr = 0; imx_epit_update_int(s); @@ -XXX,XX +XXX,XX @@ static void imx_epit_realize(DeviceState *dev, Error **errp) 0x00001000); sysbus_init_mmio(sbd, &s->iomem); + /* + * The reload timer keeps running when the peripheral is enabled. It is a + * kind of wall clock that does not generate any interrupts. The callback + * needs to be provided, but it does nothing as the ptimer already supports + * all necessary reloading functionality. + */ s->timer_reload = ptimer_init(imx_epit_reload, s, PTIMER_POLICY_LEGACY); + /* + * The compare timer is running only when the peripheral configuration is + * in a state that will generate compare interrupts. + */ s->timer_cmp = ptimer_init(imx_epit_cmp, s, PTIMER_POLICY_LEGACY); } -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> remove unused defines, add needed defines Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/timer/imx_epit.h | 4 ++-- hw/timer/imx_epit.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/include/hw/timer/imx_epit.h b/include/hw/timer/imx_epit.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/timer/imx_epit.h +++ b/include/hw/timer/imx_epit.h @@ -XXX,XX +XXX,XX @@ #define CR_OCIEN (1 << 2) #define CR_RLD (1 << 3) #define CR_PRESCALE_SHIFT (4) -#define CR_PRESCALE_MASK (0xfff) +#define CR_PRESCALE_BITS (12) #define CR_SWR (1 << 16) #define CR_IOVW (1 << 17) #define CR_DBGEN (1 << 18) @@ -XXX,XX +XXX,XX @@ #define CR_DOZEN (1 << 20) #define CR_STOPEN (1 << 21) #define CR_CLKSRC_SHIFT (24) -#define CR_CLKSRC_MASK (0x3 << CR_CLKSRC_SHIFT) +#define CR_CLKSRC_BITS (2) #define EPIT_TIMER_MAX 0XFFFFFFFFUL diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_set_freq(IMXEPITState *s) uint32_t clksrc; uint32_t prescaler; - clksrc = extract32(s->cr, CR_CLKSRC_SHIFT, 2); - prescaler = 1 + extract32(s->cr, CR_PRESCALE_SHIFT, 12); + clksrc = extract32(s->cr, CR_CLKSRC_SHIFT, CR_CLKSRC_BITS); + prescaler = 1 + extract32(s->cr, CR_PRESCALE_SHIFT, CR_PRESCALE_BITS); s->freq = imx_ccm_get_clock_frequency(s->ccm, imx_epit_clocks[clksrc]) / prescaler; -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/timer/imx_epit.h | 2 ++ hw/timer/imx_epit.c | 12 ++++++------ 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/include/hw/timer/imx_epit.h b/include/hw/timer/imx_epit.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/timer/imx_epit.h +++ b/include/hw/timer/imx_epit.h @@ -XXX,XX +XXX,XX @@ #define CR_CLKSRC_SHIFT (24) #define CR_CLKSRC_BITS (2) +#define SR_OCIF (1 << 0) + #define EPIT_TIMER_MAX 0XFFFFFFFFUL #define TYPE_IMX_EPIT "imx.epit" diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static const IMXClk imx_epit_clocks[] = { */ static void imx_epit_update_int(IMXEPITState *s) { - if (s->sr && (s->cr & CR_OCIEN) && (s->cr & CR_EN)) { + if ((s->sr & SR_OCIF) && (s->cr & CR_OCIEN) && (s->cr & CR_EN)) { qemu_irq_raise(s->irq); } else { qemu_irq_lower(s->irq); @@ -XXX,XX +XXX,XX @@ static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, break; case 1: /* SR - ACK*/ - /* writing 1 to OCIF clears the OCIF bit */ - if (value & 0x01) { - s->sr = 0; + /* writing 1 to SR.OCIF clears this bit and turns the interrupt off */ + if (value & SR_OCIF) { + s->sr = 0; /* SR.OCIF is the only bit in this register anyway */ imx_epit_update_int(s); } break; @@ -XXX,XX +XXX,XX @@ static void imx_epit_cmp(void *opaque) IMXEPITState *s = IMX_EPIT(opaque); DPRINTF("sr was %d\n", s->sr); - - s->sr = 1; + /* Set interrupt status bit SR.OCIF and update the interrupt state */ + s->sr |= SR_OCIF; imx_epit_update_int(s); } -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> The interrupt state can change due to: - reset clears both SR.OCIF and CR.OCIE - write to CR.EN or CR.OCIE Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/imx_epit.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, if (s->cr & CR_SWR) { /* handle the reset */ imx_epit_reset(DEVICE(s)); - /* - * TODO: could we 'break' here? following operations appear - * to duplicate the work imx_epit_reset() already did. - */ } + /* + * The interrupt state can change due to: + * - reset clears both SR.OCIF and CR.OCIE + * - write to CR.EN or CR.OCIE + */ + imx_epit_update_int(s); + + /* + * TODO: could we 'break' here for reset? following operations appear + * to duplicate the work imx_epit_reset() already did. + */ + ptimer_transaction_begin(s->timer_cmp); ptimer_transaction_begin(s->timer_reload); -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/imx_epit.c | 20 ++++++++++++++------ 1 file changed, 14 insertions(+), 6 deletions(-) diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_set_freq(IMXEPITState *s) /* * This is called both on hardware (device) reset and software reset. */ -static void imx_epit_reset(DeviceState *dev) +static void imx_epit_reset(IMXEPITState *s, bool is_hard_reset) { - IMXEPITState *s = IMX_EPIT(dev); - /* Soft reset doesn't touch some bits; hard reset clears them */ - s->cr &= (CR_EN|CR_ENMOD|CR_STOPEN|CR_DOZEN|CR_WAITEN|CR_DBGEN); + if (is_hard_reset) { + s->cr = 0; + } else { + s->cr &= (CR_EN|CR_ENMOD|CR_STOPEN|CR_DOZEN|CR_WAITEN|CR_DBGEN); + } s->sr = 0; s->lr = EPIT_TIMER_MAX; s->cmp = 0; @@ -XXX,XX +XXX,XX @@ static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, s->cr = value & 0x03ffffff; if (s->cr & CR_SWR) { /* handle the reset */ - imx_epit_reset(DEVICE(s)); + imx_epit_reset(s, false); } /* @@ -XXX,XX +XXX,XX @@ static void imx_epit_realize(DeviceState *dev, Error **errp) s->timer_cmp = ptimer_init(imx_epit_cmp, s, PTIMER_POLICY_LEGACY); } +static void imx_epit_dev_reset(DeviceState *dev) +{ + IMXEPITState *s = IMX_EPIT(dev); + imx_epit_reset(s, true); +} + static void imx_epit_class_init(ObjectClass *klass, void *data) { DeviceClass *dc = DEVICE_CLASS(klass); dc->realize = imx_epit_realize; - dc->reset = imx_epit_reset; + dc->reset = imx_epit_dev_reset; dc->vmsd = &vmstate_imx_timer_epit; dc->desc = "i.MX periodic timer"; } -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/imx_epit.c | 215 ++++++++++++++++++++++++-------------------- 1 file changed, 117 insertions(+), 98 deletions(-) diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_reload_compare_timer(IMXEPITState *s) } } +static void imx_epit_write_cr(IMXEPITState *s, uint32_t value) +{ + uint32_t oldcr = s->cr; + + s->cr = value & 0x03ffffff; + + if (s->cr & CR_SWR) { + /* handle the reset */ + imx_epit_reset(s, false); + } + + /* + * The interrupt state can change due to: + * - reset clears both SR.OCIF and CR.OCIE + * - write to CR.EN or CR.OCIE + */ + imx_epit_update_int(s); + + /* + * TODO: could we 'break' here for reset? following operations appear + * to duplicate the work imx_epit_reset() already did. + */ + + ptimer_transaction_begin(s->timer_cmp); + ptimer_transaction_begin(s->timer_reload); + + /* Update the frequency. Has been done already in case of a reset. */ + if (!(s->cr & CR_SWR)) { + imx_epit_set_freq(s); + } + + if (s->freq && (s->cr & CR_EN) && !(oldcr & CR_EN)) { + if (s->cr & CR_ENMOD) { + if (s->cr & CR_RLD) { + ptimer_set_limit(s->timer_reload, s->lr, 1); + ptimer_set_limit(s->timer_cmp, s->lr, 1); + } else { + ptimer_set_limit(s->timer_reload, EPIT_TIMER_MAX, 1); + ptimer_set_limit(s->timer_cmp, EPIT_TIMER_MAX, 1); + } + } + + imx_epit_reload_compare_timer(s); + ptimer_run(s->timer_reload, 0); + if (s->cr & CR_OCIEN) { + ptimer_run(s->timer_cmp, 0); + } else { + ptimer_stop(s->timer_cmp); + } + } else if (!(s->cr & CR_EN)) { + /* stop both timers */ + ptimer_stop(s->timer_reload); + ptimer_stop(s->timer_cmp); + } else if (s->cr & CR_OCIEN) { + if (!(oldcr & CR_OCIEN)) { + imx_epit_reload_compare_timer(s); + ptimer_run(s->timer_cmp, 0); + } + } else { + ptimer_stop(s->timer_cmp); + } + + ptimer_transaction_commit(s->timer_cmp); + ptimer_transaction_commit(s->timer_reload); +} + +static void imx_epit_write_sr(IMXEPITState *s, uint32_t value) +{ + /* writing 1 to SR.OCIF clears this bit and turns the interrupt off */ + if (value & SR_OCIF) { + s->sr = 0; /* SR.OCIF is the only bit in this register anyway */ + imx_epit_update_int(s); + } +} + +static void imx_epit_write_lr(IMXEPITState *s, uint32_t value) +{ + s->lr = value; + + ptimer_transaction_begin(s->timer_cmp); + ptimer_transaction_begin(s->timer_reload); + if (s->cr & CR_RLD) { + /* Also set the limit if the LRD bit is set */ + /* If IOVW bit is set then set the timer value */ + ptimer_set_limit(s->timer_reload, s->lr, s->cr & CR_IOVW); + ptimer_set_limit(s->timer_cmp, s->lr, 0); + } else if (s->cr & CR_IOVW) { + /* If IOVW bit is set then set the timer value */ + ptimer_set_count(s->timer_reload, s->lr); + } + /* + * Commit the change to s->timer_reload, so it can propagate. Otherwise + * the timer interrupt may not fire properly. The commit must happen + * before calling imx_epit_reload_compare_timer(), which reads + * s->timer_reload internally again. + */ + ptimer_transaction_commit(s->timer_reload); + imx_epit_reload_compare_timer(s); + ptimer_transaction_commit(s->timer_cmp); +} + +static void imx_epit_write_cmp(IMXEPITState *s, uint32_t value) +{ + s->cmp = value; + + ptimer_transaction_begin(s->timer_cmp); + imx_epit_reload_compare_timer(s); + ptimer_transaction_commit(s->timer_cmp); +} + static void imx_epit_write(void *opaque, hwaddr offset, uint64_t value, unsigned size) { IMXEPITState *s = IMX_EPIT(opaque); - uint64_t oldcr; DPRINTF("(%s, value = 0x%08x)\n", imx_epit_reg_name(offset >> 2), (uint32_t)value); switch (offset >> 2) { case 0: /* CR */ - - oldcr = s->cr; - s->cr = value & 0x03ffffff; - if (s->cr & CR_SWR) { - /* handle the reset */ - imx_epit_reset(s, false); - } - - /* - * The interrupt state can change due to: - * - reset clears both SR.OCIF and CR.OCIE - * - write to CR.EN or CR.OCIE - */ - imx_epit_update_int(s); - - /* - * TODO: could we 'break' here for reset? following operations appear - * to duplicate the work imx_epit_reset() already did. - */ - - ptimer_transaction_begin(s->timer_cmp); - ptimer_transaction_begin(s->timer_reload); - - /* Update the frequency. Has been done already in case of a reset. */ - if (!(s->cr & CR_SWR)) { - imx_epit_set_freq(s); - } - - if (s->freq && (s->cr & CR_EN) && !(oldcr & CR_EN)) { - if (s->cr & CR_ENMOD) { - if (s->cr & CR_RLD) { - ptimer_set_limit(s->timer_reload, s->lr, 1); - ptimer_set_limit(s->timer_cmp, s->lr, 1); - } else { - ptimer_set_limit(s->timer_reload, EPIT_TIMER_MAX, 1); - ptimer_set_limit(s->timer_cmp, EPIT_TIMER_MAX, 1); - } - } - - imx_epit_reload_compare_timer(s); - ptimer_run(s->timer_reload, 0); - if (s->cr & CR_OCIEN) { - ptimer_run(s->timer_cmp, 0); - } else { - ptimer_stop(s->timer_cmp); - } - } else if (!(s->cr & CR_EN)) { - /* stop both timers */ - ptimer_stop(s->timer_reload); - ptimer_stop(s->timer_cmp); - } else if (s->cr & CR_OCIEN) { - if (!(oldcr & CR_OCIEN)) { - imx_epit_reload_compare_timer(s); - ptimer_run(s->timer_cmp, 0); - } - } else { - ptimer_stop(s->timer_cmp); - } - - ptimer_transaction_commit(s->timer_cmp); - ptimer_transaction_commit(s->timer_reload); + imx_epit_write_cr(s, (uint32_t)value); break; - case 1: /* SR - ACK*/ - /* writing 1 to SR.OCIF clears this bit and turns the interrupt off */ - if (value & SR_OCIF) { - s->sr = 0; /* SR.OCIF is the only bit in this register anyway */ - imx_epit_update_int(s); - } + case 1: /* SR */ + imx_epit_write_sr(s, (uint32_t)value); break; - case 2: /* LR - set ticks */ - s->lr = value; - - ptimer_transaction_begin(s->timer_cmp); - ptimer_transaction_begin(s->timer_reload); - if (s->cr & CR_RLD) { - /* Also set the limit if the LRD bit is set */ - /* If IOVW bit is set then set the timer value */ - ptimer_set_limit(s->timer_reload, s->lr, s->cr & CR_IOVW); - ptimer_set_limit(s->timer_cmp, s->lr, 0); - } else if (s->cr & CR_IOVW) { - /* If IOVW bit is set then set the timer value */ - ptimer_set_count(s->timer_reload, s->lr); - } - /* - * Commit the change to s->timer_reload, so it can propagate. Otherwise - * the timer interrupt may not fire properly. The commit must happen - * before calling imx_epit_reload_compare_timer(), which reads - * s->timer_reload internally again. - */ - ptimer_transaction_commit(s->timer_reload); - imx_epit_reload_compare_timer(s); - ptimer_transaction_commit(s->timer_cmp); + case 2: /* LR */ + imx_epit_write_lr(s, (uint32_t)value); break; case 3: /* CMP */ - s->cmp = value; - - ptimer_transaction_begin(s->timer_cmp); - imx_epit_reload_compare_timer(s); - ptimer_transaction_commit(s->timer_cmp); - + imx_epit_write_cmp(s, (uint32_t)value); break; default: qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: Bad register at offset 0x%" HWADDR_PRIx "\n", TYPE_IMX_EPIT, __func__, offset); - break; } } + static void imx_epit_cmp(void *opaque) { IMXEPITState *s = IMX_EPIT(opaque); -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> The CNT register is a read-only register. There is no need to store it's value, it can be calculated on demand. The calculated frequency is needed temporarily only. Note that this is a migration compatibility break for all boards types that use the EPIT peripheral. Signed-off-by: Axel Heider <axel.heider@hensoldt.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/timer/imx_epit.h | 2 - hw/timer/imx_epit.c | 73 ++++++++++++++----------------------- 2 files changed, 28 insertions(+), 47 deletions(-) diff --git a/include/hw/timer/imx_epit.h b/include/hw/timer/imx_epit.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/timer/imx_epit.h +++ b/include/hw/timer/imx_epit.h @@ -XXX,XX +XXX,XX @@ struct IMXEPITState { uint32_t sr; uint32_t lr; uint32_t cmp; - uint32_t cnt; - uint32_t freq; qemu_irq irq; }; diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ static void imx_epit_update_int(IMXEPITState *s) } } -/* - * Must be called from within a ptimer_transaction_begin/commit block - * for both s->timer_cmp and s->timer_reload. - */ -static void imx_epit_set_freq(IMXEPITState *s) +static uint32_t imx_epit_get_freq(IMXEPITState *s) { - uint32_t clksrc; - uint32_t prescaler; - - clksrc = extract32(s->cr, CR_CLKSRC_SHIFT, CR_CLKSRC_BITS); - prescaler = 1 + extract32(s->cr, CR_PRESCALE_SHIFT, CR_PRESCALE_BITS); - - s->freq = imx_ccm_get_clock_frequency(s->ccm, - imx_epit_clocks[clksrc]) / prescaler; - - DPRINTF("Setting ptimer frequency to %u\n", s->freq); - - if (s->freq) { - ptimer_set_freq(s->timer_reload, s->freq); - ptimer_set_freq(s->timer_cmp, s->freq); - } + uint32_t clksrc = extract32(s->cr, CR_CLKSRC_SHIFT, CR_CLKSRC_BITS); + uint32_t prescaler = 1 + extract32(s->cr, CR_PRESCALE_SHIFT, CR_PRESCALE_BITS); + uint32_t f_in = imx_ccm_get_clock_frequency(s->ccm, imx_epit_clocks[clksrc]); + uint32_t freq = f_in / prescaler; + DPRINTF("ptimer frequency is %u\n", freq); + return freq; } /* @@ -XXX,XX +XXX,XX @@ static void imx_epit_reset(IMXEPITState *s, bool is_hard_reset) s->sr = 0; s->lr = EPIT_TIMER_MAX; s->cmp = 0; - s->cnt = 0; ptimer_transaction_begin(s->timer_cmp); ptimer_transaction_begin(s->timer_reload); - /* stop both timers */ + + /* + * The reset switches off the input clock, so even if the CR.EN is still + * set, the timers are no longer running. + */ + assert(imx_epit_get_freq(s) == 0); ptimer_stop(s->timer_cmp); ptimer_stop(s->timer_reload); - /* compute new frequency */ - imx_epit_set_freq(s); /* init both timers to EPIT_TIMER_MAX */ ptimer_set_limit(s->timer_cmp, EPIT_TIMER_MAX, 1); ptimer_set_limit(s->timer_reload, EPIT_TIMER_MAX, 1); - if (s->freq && (s->cr & CR_EN)) { - /* if the timer is still enabled, restart it */ - ptimer_run(s->timer_reload, 0); - } ptimer_transaction_commit(s->timer_cmp); ptimer_transaction_commit(s->timer_reload); } -static uint32_t imx_epit_update_count(IMXEPITState *s) -{ - s->cnt = ptimer_get_count(s->timer_reload); - - return s->cnt; -} - static uint64_t imx_epit_read(void *opaque, hwaddr offset, unsigned size) { IMXEPITState *s = IMX_EPIT(opaque); @@ -XXX,XX +XXX,XX @@ static uint64_t imx_epit_read(void *opaque, hwaddr offset, unsigned size) break; case 4: /* CNT */ - imx_epit_update_count(s); - reg_value = s->cnt; + reg_value = ptimer_get_count(s->timer_reload); break; default: @@ -XXX,XX +XXX,XX @@ static void imx_epit_reload_compare_timer(IMXEPITState *s) { if ((s->cr & (CR_EN | CR_OCIEN)) == (CR_EN | CR_OCIEN)) { /* if the compare feature is on and timers are running */ - uint32_t tmp = imx_epit_update_count(s); + uint32_t tmp = ptimer_get_count(s->timer_reload); uint64_t next; if (tmp > s->cmp) { /* It'll fire in this round of the timer */ @@ -XXX,XX +XXX,XX @@ static void imx_epit_reload_compare_timer(IMXEPITState *s) static void imx_epit_write_cr(IMXEPITState *s, uint32_t value) { + uint32_t freq = 0; uint32_t oldcr = s->cr; s->cr = value & 0x03ffffff; @@ -XXX,XX +XXX,XX @@ static void imx_epit_write_cr(IMXEPITState *s, uint32_t value) ptimer_transaction_begin(s->timer_cmp); ptimer_transaction_begin(s->timer_reload); - /* Update the frequency. Has been done already in case of a reset. */ + /* + * Update the frequency. In case of a reset the input clock was + * switched off, so this can be skipped. + */ if (!(s->cr & CR_SWR)) { - imx_epit_set_freq(s); + freq = imx_epit_get_freq(s); + if (freq) { + ptimer_set_freq(s->timer_reload, freq); + ptimer_set_freq(s->timer_cmp, freq); + } } - if (s->freq && (s->cr & CR_EN) && !(oldcr & CR_EN)) { + if (freq && (s->cr & CR_EN) && !(oldcr & CR_EN)) { if (s->cr & CR_ENMOD) { if (s->cr & CR_RLD) { ptimer_set_limit(s->timer_reload, s->lr, 1); @@ -XXX,XX +XXX,XX @@ static const MemoryRegionOps imx_epit_ops = { static const VMStateDescription vmstate_imx_timer_epit = { .name = TYPE_IMX_EPIT, - .version_id = 2, - .minimum_version_id = 2, + .version_id = 3, + .minimum_version_id = 3, .fields = (VMStateField[]) { VMSTATE_UINT32(cr, IMXEPITState), VMSTATE_UINT32(sr, IMXEPITState), VMSTATE_UINT32(lr, IMXEPITState), VMSTATE_UINT32(cmp, IMXEPITState), - VMSTATE_UINT32(cnt, IMXEPITState), - VMSTATE_UINT32(freq, IMXEPITState), VMSTATE_PTIMER(timer_reload, IMXEPITState), VMSTATE_PTIMER(timer_cmp, IMXEPITState), VMSTATE_END_OF_LIST() -- 2.25.1 From: Axel Heider <axel.heider@hensoldt.net> - fix #1263 for CR writes - rework compare time handling - The compare timer has to run even if CR.OCIEN is not set, as SR.OCIF must be updated. - The compare timer fires exactly once when the compare value is less than the current value, but the reload values is less than the compare value. - The compare timer will never fire if the reload value is less than the compare value. Disable it in this case. Signed-off-by: Axel Heider <axel.heider@hensoldt.net> [PMM: fixed minor style nits] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/timer/imx_epit.c | 192 ++++++++++++++++++++++++++------------------ 1 file changed, 116 insertions(+), 76 deletions(-) diff --git a/hw/timer/imx_epit.c b/hw/timer/imx_epit.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_epit.c +++ b/hw/timer/imx_epit.c @@ -XXX,XX +XXX,XX @@ * Originally written by Hans Jiang * Updated by Peter Chubb * Updated by Jean-Christophe Dubois <jcd@tribudubois.net> + * Updated by Axel Heider * * This code is licensed under GPL version 2 or later. See * the COPYING file in the top-level directory. @@ -XXX,XX +XXX,XX @@ static uint64_t imx_epit_read(void *opaque, hwaddr offset, unsigned size) return reg_value; } -/* Must be called from ptimer_transaction_begin/commit block for s->timer_cmp */ -static void imx_epit_reload_compare_timer(IMXEPITState *s) +/* + * Must be called from a ptimer_transaction_begin/commit block for + * s->timer_cmp, but outside of a transaction block of s->timer_reload, + * so the proper counter value is read. + */ +static void imx_epit_update_compare_timer(IMXEPITState *s) { - if ((s->cr & (CR_EN | CR_OCIEN)) == (CR_EN | CR_OCIEN)) { - /* if the compare feature is on and timers are running */ - uint32_t tmp = ptimer_get_count(s->timer_reload); - uint64_t next; - if (tmp > s->cmp) { - /* It'll fire in this round of the timer */ - next = tmp - s->cmp; - } else { /* catch it next time around */ - next = tmp - s->cmp + ((s->cr & CR_RLD) ? EPIT_TIMER_MAX : s->lr); + uint64_t counter = 0; + bool is_oneshot = false; + /* + * The compare timer only has to run if the timer peripheral is active + * and there is an input clock, Otherwise it can be switched off. + */ + bool is_active = (s->cr & CR_EN) && imx_epit_get_freq(s); + if (is_active) { + /* + * Calculate next timeout for compare timer. Reading the reload + * counter returns proper results only if pending transactions + * on it are committed here. Otherwise stale values are be read. + */ + counter = ptimer_get_count(s->timer_reload); + uint64_t limit = ptimer_get_limit(s->timer_cmp); + /* + * The compare timer is a periodic timer if the limit is at least + * the compare value. Otherwise it may fire at most once in the + * current round. + */ + bool is_oneshot = (limit >= s->cmp); + if (counter >= s->cmp) { + /* The compare timer fires in the current round. */ + counter -= s->cmp; + } else if (!is_oneshot) { + /* + * The compare timer fires after a reload, as it is below the + * compare value already in this round. Note that the counter + * value calculated below can be above the 32-bit limit, which + * is legal here because the compare timer is an internal + * helper ptimer only. + */ + counter += limit - s->cmp; + } else { + /* + * The compare timer won't fire in this round, and the limit is + * set to a value below the compare value. This practically means + * it will never fire, so it can be switched off. + */ + is_active = false; } - ptimer_set_count(s->timer_cmp, next); } + + /* + * Set the compare timer and let it run, or stop it. This is agnostic + * of CR.OCIEN bit, as this bit affects interrupt generation only. The + * compare timer needs to run even if no interrupts are to be generated, + * because the SR.OCIF bit must be updated also. + * Note that the timer might already be stopped or be running with + * counter values. However, finding out when an update is needed and + * when not is not trivial. It's much easier applying the setting again, + * as this does not harm either and the overhead is negligible. + */ + if (is_active) { + ptimer_set_count(s->timer_cmp, counter); + ptimer_run(s->timer_cmp, is_oneshot ? 1 : 0); + } else { + ptimer_stop(s->timer_cmp); + } + } static void imx_epit_write_cr(IMXEPITState *s, uint32_t value) { - uint32_t freq = 0; uint32_t oldcr = s->cr; s->cr = value & 0x03ffffff; if (s->cr & CR_SWR) { - /* handle the reset */ + /* + * Reset clears CR.SWR again. It does not touch CR.EN, but the timers + * are still stopped because the input clock is disabled. + */ imx_epit_reset(s, false); + } else { + uint32_t freq; + uint32_t toggled_cr_bits = oldcr ^ s->cr; + /* re-initialize the limits if CR.RLD has changed */ + bool set_limit = toggled_cr_bits & CR_RLD; + /* set the counter if the timer got just enabled and CR.ENMOD is set */ + bool is_switched_on = (toggled_cr_bits & s->cr) & CR_EN; + bool set_counter = is_switched_on && (s->cr & CR_ENMOD); + + ptimer_transaction_begin(s->timer_cmp); + ptimer_transaction_begin(s->timer_reload); + freq = imx_epit_get_freq(s); + if (freq) { + ptimer_set_freq(s->timer_reload, freq); + ptimer_set_freq(s->timer_cmp, freq); + } + + if (set_limit || set_counter) { + uint64_t limit = (s->cr & CR_RLD) ? s->lr : EPIT_TIMER_MAX; + ptimer_set_limit(s->timer_reload, limit, set_counter ? 1 : 0); + if (set_limit) { + ptimer_set_limit(s->timer_cmp, limit, 0); + } + } + /* + * If there is an input clock and the peripheral is enabled, then + * ensure the wall clock timer is ticking. Otherwise stop the timers. + * The compare timer will be updated later. + */ + if (freq && (s->cr & CR_EN)) { + ptimer_run(s->timer_reload, 0); + } else { + ptimer_stop(s->timer_reload); + } + /* Commit changes to reload timer, so they can propagate. */ + ptimer_transaction_commit(s->timer_reload); + /* Update compare timer based on the committed reload timer value. */ + imx_epit_update_compare_timer(s); + ptimer_transaction_commit(s->timer_cmp); } /* @@ -XXX,XX +XXX,XX @@ static void imx_epit_write_cr(IMXEPITState *s, uint32_t value) * - write to CR.EN or CR.OCIE */ imx_epit_update_int(s); - - /* - * TODO: could we 'break' here for reset? following operations appear - * to duplicate the work imx_epit_reset() already did. - */ - - ptimer_transaction_begin(s->timer_cmp); - ptimer_transaction_begin(s->timer_reload); - - /* - * Update the frequency. In case of a reset the input clock was - * switched off, so this can be skipped. - */ - if (!(s->cr & CR_SWR)) { - freq = imx_epit_get_freq(s); - if (freq) { - ptimer_set_freq(s->timer_reload, freq); - ptimer_set_freq(s->timer_cmp, freq); - } - } - - if (freq && (s->cr & CR_EN) && !(oldcr & CR_EN)) { - if (s->cr & CR_ENMOD) { - if (s->cr & CR_RLD) { - ptimer_set_limit(s->timer_reload, s->lr, 1); - ptimer_set_limit(s->timer_cmp, s->lr, 1); - } else { - ptimer_set_limit(s->timer_reload, EPIT_TIMER_MAX, 1); - ptimer_set_limit(s->timer_cmp, EPIT_TIMER_MAX, 1); - } - } - - imx_epit_reload_compare_timer(s); - ptimer_run(s->timer_reload, 0); - if (s->cr & CR_OCIEN) { - ptimer_run(s->timer_cmp, 0); - } else { - ptimer_stop(s->timer_cmp); - } - } else if (!(s->cr & CR_EN)) { - /* stop both timers */ - ptimer_stop(s->timer_reload); - ptimer_stop(s->timer_cmp); - } else if (s->cr & CR_OCIEN) { - if (!(oldcr & CR_OCIEN)) { - imx_epit_reload_compare_timer(s); - ptimer_run(s->timer_cmp, 0); - } - } else { - ptimer_stop(s->timer_cmp); - } - - ptimer_transaction_commit(s->timer_cmp); - ptimer_transaction_commit(s->timer_reload); } static void imx_epit_write_sr(IMXEPITState *s, uint32_t value) @@ -XXX,XX +XXX,XX @@ static void imx_epit_write_lr(IMXEPITState *s, uint32_t value) /* If IOVW bit is set then set the timer value */ ptimer_set_count(s->timer_reload, s->lr); } - /* - * Commit the change to s->timer_reload, so it can propagate. Otherwise - * the timer interrupt may not fire properly. The commit must happen - * before calling imx_epit_reload_compare_timer(), which reads - * s->timer_reload internally again. - */ + /* Commit the changes to s->timer_reload, so they can propagate. */ ptimer_transaction_commit(s->timer_reload); - imx_epit_reload_compare_timer(s); + /* Update the compare timer based on the committed reload timer value. */ + imx_epit_update_compare_timer(s); ptimer_transaction_commit(s->timer_cmp); } @@ -XXX,XX +XXX,XX @@ static void imx_epit_write_cmp(IMXEPITState *s, uint32_t value) { s->cmp = value; + /* Update the compare timer based on the committed reload timer value. */ ptimer_transaction_begin(s->timer_cmp); - imx_epit_reload_compare_timer(s); + imx_epit_update_compare_timer(s); ptimer_transaction_commit(s->timer_cmp); } @@ -XXX,XX +XXX,XX @@ static void imx_epit_cmp(void *opaque) { IMXEPITState *s = IMX_EPIT(opaque); + /* The cmp ptimer can't be running when the peripheral is disabled */ + assert(s->cr & CR_EN); + DPRINTF("sr was %d\n", s->sr); /* Set interrupt status bit SR.OCIF and update the interrupt state */ s->sr |= SR_OCIF; -- 2.25.1 From: Fabiano Rosas <farosas@suse.de> Fix these: WARNING: Block comments use a leading /* on a separate line WARNING: Block comments use * on subsequent lines WARNING: Block comments use a trailing */ on a separate line Signed-off-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Message-id: 20221213190537.511-2-farosas@suse.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 323 +++++++++++++++++++++++++++++--------------- 1 file changed, 215 insertions(+), 108 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ uint64_t read_raw_cp_reg(CPUARMState *env, const ARMCPRegInfo *ri) static void write_raw_cp_reg(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t v) { - /* Raw write of a coprocessor register (as needed for migration, etc). + /* + * Raw write of a coprocessor register (as needed for migration, etc). * Note that constant registers are treated as write-ignored; the * caller should check for success by whether a readback gives the * value written. @@ -XXX,XX +XXX,XX @@ static void write_raw_cp_reg(CPUARMState *env, const ARMCPRegInfo *ri, static bool raw_accessors_invalid(const ARMCPRegInfo *ri) { - /* Return true if the regdef would cause an assertion if you called + /* + * Return true if the regdef would cause an assertion if you called * read_raw_cp_reg() or write_raw_cp_reg() on it (ie if it is a * program bug for it not to have the NO_RAW flag). * NB that returning false here doesn't necessarily mean that calling @@ -XXX,XX +XXX,XX @@ bool write_list_to_cpustate(ARMCPU *cpu) if (ri->type & ARM_CP_NO_RAW) { continue; } - /* Write value and confirm it reads back as written + /* + * Write value and confirm it reads back as written * (to catch read-only registers and partially read-only * registers where the incoming migration value doesn't match) */ @@ -XXX,XX +XXX,XX @@ static gint cpreg_key_compare(gconstpointer a, gconstpointer b) void init_cpreg_list(ARMCPU *cpu) { - /* Initialise the cpreg_tuples[] array based on the cp_regs hash. + /* + * Initialise the cpreg_tuples[] array based on the cp_regs hash. * Note that we require cpreg_tuples[] to be sorted by key ID. */ GList *keys; @@ -XXX,XX +XXX,XX @@ static CPAccessResult access_el3_aa32ns(CPUARMState *env, return CP_ACCESS_OK; } -/* Some secure-only AArch32 registers trap to EL3 if used from +/* + * Some secure-only AArch32 registers trap to EL3 if used from * Secure EL1 (but are just ordinary UNDEF in other non-EL3 contexts). * Note that an access from Secure EL1 can only happen if EL3 is AArch64. * We assume that the .access field is set to PL1_RW. @@ -XXX,XX +XXX,XX @@ static CPAccessResult access_trap_aa32s_el1(CPUARMState *env, return CP_ACCESS_TRAP_UNCATEGORIZED; } -/* Check for traps to performance monitor registers, which are controlled +/* + * Check for traps to performance monitor registers, which are controlled * by MDCR_EL2.TPM for EL2 and MDCR_EL3.TPM for EL3. */ static CPAccessResult access_tpm(CPUARMState *env, const ARMCPRegInfo *ri, @@ -XXX,XX +XXX,XX @@ static void fcse_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) ARMCPU *cpu = env_archcpu(env); if (raw_read(env, ri) != value) { - /* Unlike real hardware the qemu TLB uses virtual addresses, + /* + * Unlike real hardware the qemu TLB uses virtual addresses, * not modified virtual addresses, so this causes a TLB flush. */ tlb_flush(CPU(cpu)); @@ -XXX,XX +XXX,XX @@ static void contextidr_write(CPUARMState *env, const ARMCPRegInfo *ri, if (raw_read(env, ri) != value && !arm_feature(env, ARM_FEATURE_PMSA) && !extended_addresses_enabled(env)) { - /* For VMSA (when not using the LPAE long descriptor page table + /* + * For VMSA (when not using the LPAE long descriptor page table * format) this register includes the ASID, so do a TLB flush. * For PMSA it is purely a process ID and no action is needed. */ @@ -XXX,XX +XXX,XX @@ static void tlbiipas2is_hyp_write(CPUARMState *env, const ARMCPRegInfo *ri, } static const ARMCPRegInfo cp_reginfo[] = { - /* Define the secure and non-secure FCSE identifier CP registers + /* + * Define the secure and non-secure FCSE identifier CP registers * separately because there is no secure bank in V8 (no _EL3). This allows * the secure register to be properly reset and migrated. There is also no * v8 EL1 version of the register so the non-secure instance stands alone. @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo cp_reginfo[] = { .access = PL1_RW, .secure = ARM_CP_SECSTATE_S, .fieldoffset = offsetof(CPUARMState, cp15.fcseidr_s), .resetvalue = 0, .writefn = fcse_write, .raw_writefn = raw_write, }, - /* Define the secure and non-secure context identifier CP registers + /* + * Define the secure and non-secure context identifier CP registers * separately because there is no secure bank in V8 (no _EL3). This allows * the secure register to be properly reset and migrated. In the * non-secure case, the 32-bit register will have reset and migration @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo cp_reginfo[] = { }; static const ARMCPRegInfo not_v8_cp_reginfo[] = { - /* NB: Some of these registers exist in v8 but with more precise + /* + * NB: Some of these registers exist in v8 but with more precise * definitions that don't use CP_ANY wildcards (mostly in v8_cp_reginfo[]). */ /* MMU Domain access control / MPU write buffer control */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo not_v8_cp_reginfo[] = { .writefn = dacr_write, .raw_writefn = raw_write, .bank_fieldoffsets = { offsetoflow32(CPUARMState, cp15.dacr_s), offsetoflow32(CPUARMState, cp15.dacr_ns) } }, - /* ARMv7 allocates a range of implementation defined TLB LOCKDOWN regs. + /* + * ARMv7 allocates a range of implementation defined TLB LOCKDOWN regs. * For v6 and v5, these mappings are overly broad. */ { .name = "TLB_LOCKDOWN", .cp = 15, .crn = 10, .crm = 0, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo not_v8_cp_reginfo[] = { }; static const ARMCPRegInfo not_v6_cp_reginfo[] = { - /* Not all pre-v6 cores implemented this WFI, so this is slightly + /* + * Not all pre-v6 cores implemented this WFI, so this is slightly * over-broad. */ { .name = "WFI_v5", .cp = 15, .crn = 7, .crm = 8, .opc1 = 0, .opc2 = 2, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo not_v6_cp_reginfo[] = { }; static const ARMCPRegInfo not_v7_cp_reginfo[] = { - /* Standard v6 WFI (also used in some pre-v6 cores); not in v7 (which + /* + * Standard v6 WFI (also used in some pre-v6 cores); not in v7 (which * is UNPREDICTABLE; we choose to NOP as most implementations do). */ { .name = "WFI_v6", .cp = 15, .crn = 7, .crm = 0, .opc1 = 0, .opc2 = 4, .access = PL1_W, .type = ARM_CP_WFI }, - /* L1 cache lockdown. Not architectural in v6 and earlier but in practice + /* + * L1 cache lockdown. Not architectural in v6 and earlier but in practice * implemented in 926, 946, 1026, 1136, 1176 and 11MPCore. StrongARM and * OMAPCP will override this space. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo not_v7_cp_reginfo[] = { { .name = "DUMMY", .cp = 15, .crn = 0, .crm = 0, .opc1 = 1, .opc2 = CP_ANY, .access = PL1_R, .type = ARM_CP_CONST | ARM_CP_NO_RAW, .resetvalue = 0 }, - /* We don't implement pre-v7 debug but most CPUs had at least a DBGDIDR; + /* + * We don't implement pre-v7 debug but most CPUs had at least a DBGDIDR; * implementing it as RAZ means the "debug architecture version" bits * will read as a reserved value, which should cause Linux to not try * to use the debug hardware. */ { .name = "DBGDIDR", .cp = 14, .crn = 0, .crm = 0, .opc1 = 0, .opc2 = 0, .access = PL0_R, .type = ARM_CP_CONST, .resetvalue = 0 }, - /* MMU TLB control. Note that the wildcarding means we cover not just + /* + * MMU TLB control. Note that the wildcarding means we cover not just * the unified TLB ops but also the dside/iside/inner-shareable variants. */ { .name = "TLBIALL", .cp = 15, .crn = 8, .crm = CP_ANY, @@ -XXX,XX +XXX,XX @@ static void cpacr_write(CPUARMState *env, const ARMCPRegInfo *ri, /* In ARMv8 most bits of CPACR_EL1 are RES0. */ if (!arm_feature(env, ARM_FEATURE_V8)) { - /* ARMv7 defines bits for unimplemented coprocessors as RAZ/WI. + /* + * ARMv7 defines bits for unimplemented coprocessors as RAZ/WI. * ASEDIS [31] and D32DIS [30] are both UNK/SBZP without VFP. * TRCDIS [28] is RAZ/WI since we do not implement a trace macrocell. */ @@ -XXX,XX +XXX,XX @@ static void cpacr_write(CPUARMState *env, const ARMCPRegInfo *ri, value |= R_CPACR_ASEDIS_MASK; } - /* VFPv3 and upwards with NEON implement 32 double precision + /* + * VFPv3 and upwards with NEON implement 32 double precision * registers (D0-D31). */ if (!cpu_isar_feature(aa32_simd_r32, env_archcpu(env))) { @@ -XXX,XX +XXX,XX @@ static uint64_t cpacr_read(CPUARMState *env, const ARMCPRegInfo *ri) static void cpacr_reset(CPUARMState *env, const ARMCPRegInfo *ri) { - /* Call cpacr_write() so that we reset with the correct RAO bits set + /* + * Call cpacr_write() so that we reset with the correct RAO bits set * for our CPU features. */ cpacr_write(env, ri, 0); @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v6_cp_reginfo[] = { { .name = "MVA_prefetch", .cp = 15, .crn = 7, .crm = 13, .opc1 = 0, .opc2 = 1, .access = PL1_W, .type = ARM_CP_NOP }, - /* We need to break the TB after ISB to execute self-modifying code + /* + * We need to break the TB after ISB to execute self-modifying code * correctly and also to take any pending interrupts immediately. * So use arm_cp_write_ignore() function instead of ARM_CP_NOP flag. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v6_cp_reginfo[] = { .bank_fieldoffsets = { offsetof(CPUARMState, cp15.ifar_s), offsetof(CPUARMState, cp15.ifar_ns) }, .resetvalue = 0, }, - /* Watchpoint Fault Address Register : should actually only be present + /* + * Watchpoint Fault Address Register : should actually only be present * for 1136, 1176, 11MPCore. */ { .name = "WFAR", .cp = 15, .crn = 6, .crm = 0, .opc1 = 0, .opc2 = 1, @@ -XXX,XX +XXX,XX @@ static bool event_supported(uint16_t number) static CPAccessResult pmreg_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { - /* Performance monitor registers user accessibility is controlled + /* + * Performance monitor registers user accessibility is controlled * by PMUSERENR. MDCR_EL2.TPM and MDCR_EL3.TPM allow configurable * trapping to EL2 or EL3 for other accesses. */ @@ -XXX,XX +XXX,XX @@ static CPAccessResult pmreg_access_ccntr(CPUARMState *env, (MDCR_HPME | MDCR_HPMD | MDCR_HPMN | MDCR_HCCD | MDCR_HLP) #define MDCR_EL3_PMU_ENABLE_BITS (MDCR_SPME | MDCR_SCCD) -/* Returns true if the counter (pass 31 for PMCCNTR) should count events using +/* + * Returns true if the counter (pass 31 for PMCCNTR) should count events using * the current EL, security state, and register configuration. */ static bool pmu_counter_enabled(CPUARMState *env, uint8_t counter) @@ -XXX,XX +XXX,XX @@ static uint64_t pmccntr_read(CPUARMState *env, const ARMCPRegInfo *ri) static void pmselr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* The value of PMSELR.SEL affects the behavior of PMXEVTYPER and + /* + * The value of PMSELR.SEL affects the behavior of PMXEVTYPER and * PMXEVCNTR. We allow [0..31] to be written to PMSELR here; in the * meanwhile, we check PMSELR.SEL when PMXEVTYPER and PMXEVCNTR are * accessed. @@ -XXX,XX +XXX,XX @@ static void pmevtyper_write(CPUARMState *env, const ARMCPRegInfo *ri, env->cp15.c14_pmevtyper[counter] = value & PMXEVTYPER_MASK; pmevcntr_op_finish(env, counter); } - /* Attempts to access PMXEVTYPER are CONSTRAINED UNPREDICTABLE when + /* + * Attempts to access PMXEVTYPER are CONSTRAINED UNPREDICTABLE when * PMSELR value is equal to or greater than the number of implemented * counters, but not equal to 0x1f. We opt to behave as a RAZ/WI. */ @@ -XXX,XX +XXX,XX @@ static uint64_t pmevcntr_read(CPUARMState *env, const ARMCPRegInfo *ri, } return ret; } else { - /* We opt to behave as a RAZ/WI when attempts to access PM[X]EVCNTR - * are CONSTRAINED UNPREDICTABLE. */ + /* + * We opt to behave as a RAZ/WI when attempts to access PM[X]EVCNTR + * are CONSTRAINED UNPREDICTABLE. + */ return 0; } } @@ -XXX,XX +XXX,XX @@ static void pmintenclr_write(CPUARMState *env, const ARMCPRegInfo *ri, static void vbar_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* Note that even though the AArch64 view of this register has bits + /* + * Note that even though the AArch64 view of this register has bits * [10:0] all RES0 we can only mask the bottom 5, to comply with the * architectural requirements for bits which are RES0 only in some * contexts. (ARMv8 would permit us to do no masking at all, but ARMv7 @@ -XXX,XX +XXX,XX @@ static void scr_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) if (!arm_feature(env, ARM_FEATURE_EL2)) { valid_mask &= ~SCR_HCE; - /* On ARMv7, SMD (or SCD as it is called in v7) is only + /* + * On ARMv7, SMD (or SCD as it is called in v7) is only * supported if EL2 exists. The bit is UNK/SBZP when * EL2 is unavailable. In QEMU ARMv7, we force it to always zero * when EL2 is unavailable. @@ -XXX,XX +XXX,XX @@ static uint64_t ccsidr_read(CPUARMState *env, const ARMCPRegInfo *ri) { ARMCPU *cpu = env_archcpu(env); - /* Acquire the CSSELR index from the bank corresponding to the CCSIDR + /* + * Acquire the CSSELR index from the bank corresponding to the CCSIDR * bank */ uint32_t index = A32_BANKED_REG_GET(env, csselr, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { /* the old v6 WFI, UNPREDICTABLE in v7 but we choose to NOP */ { .name = "NOP", .cp = 15, .crn = 7, .crm = 0, .opc1 = 0, .opc2 = 4, .access = PL1_W, .type = ARM_CP_NOP }, - /* Performance monitors are implementation defined in v7, + /* + * Performance monitors are implementation defined in v7, * but with an ARM recommended set of registers, which we * follow. * @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .writefn = csselr_write, .resetvalue = 0, .bank_fieldoffsets = { offsetof(CPUARMState, cp15.csselr_s), offsetof(CPUARMState, cp15.csselr_ns) } }, - /* Auxiliary ID register: this actually has an IMPDEF value but for now + /* + * Auxiliary ID register: this actually has an IMPDEF value but for now * just RAZ for all cores: */ { .name = "AIDR", .state = ARM_CP_STATE_BOTH, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .access = PL1_R, .type = ARM_CP_CONST, .accessfn = access_aa64_tid1, .resetvalue = 0 }, - /* Auxiliary fault status registers: these also are IMPDEF, and we + /* + * Auxiliary fault status registers: these also are IMPDEF, and we * choose to RAZ/WI for all cores. */ { .name = "AFSR0_EL1", .state = ARM_CP_STATE_BOTH, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .opc0 = 3, .opc1 = 0, .crn = 5, .crm = 1, .opc2 = 1, .access = PL1_RW, .accessfn = access_tvm_trvm, .type = ARM_CP_CONST, .resetvalue = 0 }, - /* MAIR can just read-as-written because we don't implement caches + /* + * MAIR can just read-as-written because we don't implement caches * and so don't need to care about memory attributes. */ { .name = "MAIR_EL1", .state = ARM_CP_STATE_AA64, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v7_cp_reginfo[] = { .opc0 = 3, .opc1 = 6, .crn = 10, .crm = 2, .opc2 = 0, .access = PL3_RW, .fieldoffset = offsetof(CPUARMState, cp15.mair_el[3]), .resetvalue = 0 }, - /* For non-long-descriptor page tables these are PRRR and NMRR; + /* + * For non-long-descriptor page tables these are PRRR and NMRR; * regardless they still act as reads-as-written for QEMU. */ - /* MAIR0/1 are defined separately from their 64-bit counterpart which + /* + * MAIR0/1 are defined separately from their 64-bit counterpart which * allows them to assign the correct fieldoffset based on the endianness * handled in the field definitions. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v6k_cp_reginfo[] = { static CPAccessResult gt_cntfrq_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { - /* CNTFRQ: not visible from PL0 if both PL0PCTEN and PL0VCTEN are zero. + /* + * CNTFRQ: not visible from PL0 if both PL0PCTEN and PL0VCTEN are zero. * Writable only at the highest implemented exception level. */ int el = arm_current_el(env); @@ -XXX,XX +XXX,XX @@ static CPAccessResult gt_stimer_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { - /* The AArch64 register view of the secure physical timer is + /* + * The AArch64 register view of the secure physical timer is * always accessible from EL3, and configurably accessible from * Secure EL1. */ @@ -XXX,XX +XXX,XX @@ static void gt_recalc_timer(ARMCPU *cpu, int timeridx) ARMGenericTimer *gt = &cpu->env.cp15.c14_timer[timeridx]; if (gt->ctl & 1) { - /* Timer enabled: calculate and set current ISTATUS, irq, and + /* + * Timer enabled: calculate and set current ISTATUS, irq, and * reset timer to when ISTATUS next has to change */ uint64_t offset = timeridx == GTIMER_VIRT ? @@ -XXX,XX +XXX,XX @@ static void gt_recalc_timer(ARMCPU *cpu, int timeridx) /* Next transition is when we hit cval */ nexttick = gt->cval + offset; } - /* Note that the desired next expiry time might be beyond the + /* + * Note that the desired next expiry time might be beyond the * signed-64-bit range of a QEMUTimer -- in this case we just * set the timer for as far in the future as possible. When the * timer expires we will reset the timer for any remaining period. @@ -XXX,XX +XXX,XX @@ static void gt_ctl_write(CPUARMState *env, const ARMCPRegInfo *ri, /* Enable toggled */ gt_recalc_timer(cpu, timeridx); } else if ((oldval ^ value) & 2) { - /* IMASK toggled: don't need to recalculate, + /* + * IMASK toggled: don't need to recalculate, * just set the interrupt line based on ISTATUS */ int irqstate = (oldval & 4) && !(value & 2); @@ -XXX,XX +XXX,XX @@ static void arm_gt_cntfrq_reset(CPUARMState *env, const ARMCPRegInfo *opaque) } static const ARMCPRegInfo generic_timer_cp_reginfo[] = { - /* Note that CNTFRQ is purely reads-as-written for the benefit + /* + * Note that CNTFRQ is purely reads-as-written for the benefit * of software; writing it doesn't actually change the timer frequency. * Our reset value matches the fixed frequency we implement the timer at. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo generic_timer_cp_reginfo[] = { .readfn = gt_virt_redir_cval_read, .raw_readfn = raw_read, .writefn = gt_virt_redir_cval_write, .raw_writefn = raw_write, }, - /* Secure timer -- this is actually restricted to only EL3 + /* + * Secure timer -- this is actually restricted to only EL3 * and configurably Secure-EL1 via the accessfn. */ { .name = "CNTPS_TVAL_EL1", .state = ARM_CP_STATE_AA64, @@ -XXX,XX +XXX,XX @@ static CPAccessResult e2h_access(CPUARMState *env, const ARMCPRegInfo *ri, #else -/* In user-mode most of the generic timer registers are inaccessible +/* + * In user-mode most of the generic timer registers are inaccessible * however modern kernels (4.12+) allow access to cntvct_el0 */ @@ -XXX,XX +XXX,XX @@ static uint64_t gt_virt_cnt_read(CPUARMState *env, const ARMCPRegInfo *ri) { ARMCPU *cpu = env_archcpu(env); - /* Currently we have no support for QEMUTimer in linux-user so we + /* + * Currently we have no support for QEMUTimer in linux-user so we * can't call gt_get_countervalue(env), instead we directly * call the lower level functions. */ @@ -XXX,XX +XXX,XX @@ static CPAccessResult ats_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { if (ri->opc2 & 4) { - /* The ATS12NSO* operations must trap to EL3 or EL2 if executed in + /* + * The ATS12NSO* operations must trap to EL3 or EL2 if executed in * Secure EL1 (which can only happen if EL3 is AArch64). * They are simply UNDEF if executed from NS EL1. * They function normally from EL2 or EL3. @@ -XXX,XX +XXX,XX @@ static uint64_t do_ats_write(CPUARMState *env, uint64_t value, } } } else { - /* fsr is a DFSR/IFSR value for the short descriptor + /* + * fsr is a DFSR/IFSR value for the short descriptor * translation table format (with WnR always clear). * Convert it to a 32-bit PAR. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo pmsav8r_cp_reginfo[] = { }; static const ARMCPRegInfo pmsav7_cp_reginfo[] = { - /* Reset for all these registers is handled in arm_cpu_reset(), + /* + * Reset for all these registers is handled in arm_cpu_reset(), * because the PMSAv7 is also used by M-profile CPUs, which do * not register cpregs but still need the state to be reset. */ @@ -XXX,XX +XXX,XX @@ static void vmsa_ttbcr_write(CPUARMState *env, const ARMCPRegInfo *ri, } if (arm_feature(env, ARM_FEATURE_LPAE)) { - /* With LPAE the TTBCR could result in a change of ASID + /* + * With LPAE the TTBCR could result in a change of ASID * via the TTBCR.A1 bit, so do a TLB flush. */ tlb_flush(CPU(cpu)); @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo vmsa_cp_reginfo[] = { offsetoflow32(CPUARMState, cp15.tcr_el[1])} }, }; -/* Note that unlike TTBCR, writing to TTBCR2 does not require flushing +/* + * Note that unlike TTBCR, writing to TTBCR2 does not require flushing * qemu tlbs nor adjusting cached masks. */ static const ARMCPRegInfo ttbcr2_reginfo = { @@ -XXX,XX +XXX,XX @@ static void omap_wfi_write(CPUARMState *env, const ARMCPRegInfo *ri, static void omap_cachemaint_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* On OMAP there are registers indicating the max/min index of dcache lines + /* + * On OMAP there are registers indicating the max/min index of dcache lines * containing a dirty line; cache flush operations have to reset these. */ env->cp15.c15_i_max = 0x000; @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo omap_cp_reginfo[] = { .crm = 8, .opc1 = 0, .opc2 = 0, .access = PL1_RW, .type = ARM_CP_NO_RAW, .readfn = arm_cp_read_zero, .writefn = omap_wfi_write, }, - /* TODO: Peripheral port remap register: + /* + * TODO: Peripheral port remap register: * On OMAP2 mcr p15, 0, rn, c15, c2, 4 sets up the interrupt controller * base address at $rn & ~0xfff and map size of 0x200 << ($rn & 0xfff), * when MMU is off. @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo xscale_cp_reginfo[] = { .cp = 15, .crn = 1, .crm = 0, .opc1 = 0, .opc2 = 1, .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, cp15.c1_xscaleauxcr), .resetvalue = 0, }, - /* XScale specific cache-lockdown: since we have no cache we NOP these + /* + * XScale specific cache-lockdown: since we have no cache we NOP these * and hope the guest does not really rely on cache behaviour. */ { .name = "XSCALE_LOCK_ICACHE_LINE", @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo xscale_cp_reginfo[] = { }; static const ARMCPRegInfo dummy_c15_cp_reginfo[] = { - /* RAZ/WI the whole crn=15 space, when we don't have a more specific + /* + * RAZ/WI the whole crn=15 space, when we don't have a more specific * implementation of this implementation-defined space. * Ideally this should eventually disappear in favour of actually * implementing the correct behaviour for all cores. @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo cache_block_ops_cp_reginfo[] = { }; static const ARMCPRegInfo cache_test_clean_cp_reginfo[] = { - /* The cache test-and-clean instructions always return (1 << 30) + /* + * The cache test-and-clean instructions always return (1 << 30) * to indicate that there are no dirty cache lines. */ { .name = "TC_DCACHE", .cp = 15, .crn = 7, .crm = 10, .opc1 = 0, .opc2 = 3, @@ -XXX,XX +XXX,XX @@ static uint64_t mpidr_read_val(CPUARMState *env) if (arm_feature(env, ARM_FEATURE_V7MP)) { mpidr |= (1U << 31); - /* Cores which are uniprocessor (non-coherent) + /* + * Cores which are uniprocessor (non-coherent) * but still implement the MP extensions set * bit 30. (For instance, Cortex-R5). */ @@ -XXX,XX +XXX,XX @@ static CPAccessResult access_tocu(CPUARMState *env, const ARMCPRegInfo *ri, return do_cacheop_pou_access(env, HCR_TOCU | HCR_TPU); } -/* See: D4.7.2 TLB maintenance requirements and the TLB maintenance instructions +/* + * See: D4.7.2 TLB maintenance requirements and the TLB maintenance instructions * Page D4-1736 (DDI0487A.b) */ @@ -XXX,XX +XXX,XX @@ static void tlbi_aa64_alle3is_write(CPUARMState *env, const ARMCPRegInfo *ri, static void tlbi_aa64_vae2_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* Invalidate by VA, EL2 + /* + * Invalidate by VA, EL2 * Currently handles both VAE2 and VALE2, since we don't support * flush-last-level-only. */ @@ -XXX,XX +XXX,XX @@ static void tlbi_aa64_vae2_write(CPUARMState *env, const ARMCPRegInfo *ri, static void tlbi_aa64_vae3_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* Invalidate by VA, EL3 + /* + * Invalidate by VA, EL3 * Currently handles both VAE3 and VALE3, since we don't support * flush-last-level-only. */ @@ -XXX,XX +XXX,XX @@ static void tlbi_aa64_vae1is_write(CPUARMState *env, const ARMCPRegInfo *ri, static void tlbi_aa64_vae1_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) { - /* Invalidate by VA, EL1&0 (AArch64 version). + /* + * Invalidate by VA, EL1&0 (AArch64 version). * Currently handles all of VAE1, VAAE1, VAALE1 and VALE1, * since we don't support flush-for-specific-ASID-only or * flush-last-level-only. @@ -XXX,XX +XXX,XX @@ static CPAccessResult sp_el0_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { if (!(env->pstate & PSTATE_SP)) { - /* Access to SP_EL0 is undefined if it's being used as + /* + * Access to SP_EL0 is undefined if it's being used as * the stack pointer. */ return CP_ACCESS_TRAP_UNCATEGORIZED; @@ -XXX,XX +XXX,XX @@ static void sctlr_write(CPUARMState *env, const ARMCPRegInfo *ri, } if (raw_read(env, ri) == value) { - /* Skip the TLB flush if nothing actually changed; Linux likes + /* + * Skip the TLB flush if nothing actually changed; Linux likes * to do a lot of pointless SCTLR writes. */ return; @@ -XXX,XX +XXX,XX @@ static void mdcr_el2_write(CPUARMState *env, const ARMCPRegInfo *ri, } static const ARMCPRegInfo v8_cp_reginfo[] = { - /* Minimal set of EL0-visible registers. This will need to be expanded + /* + * Minimal set of EL0-visible registers. This will need to be expanded * significantly for system emulation of AArch64 CPUs. */ { .name = "NZCV", .state = ARM_CP_STATE_AA64, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v8_cp_reginfo[] = { .opc0 = 3, .opc1 = 0, .crn = 4, .crm = 0, .opc2 = 0, .access = PL1_RW, .fieldoffset = offsetof(CPUARMState, banked_spsr[BANK_SVC]) }, - /* We rely on the access checks not allowing the guest to write to the + /* + * We rely on the access checks not allowing the guest to write to the * state field when SPSel indicates that it's being used as the stack * pointer. */ @@ -XXX,XX +XXX,XX @@ static void do_hcr_write(CPUARMState *env, uint64_t value, uint64_t valid_mask) if (arm_feature(env, ARM_FEATURE_EL3)) { valid_mask &= ~HCR_HCD; } else if (cpu->psci_conduit != QEMU_PSCI_CONDUIT_SMC) { - /* Architecturally HCR.TSC is RES0 if EL3 is not implemented. + /* + * Architecturally HCR.TSC is RES0 if EL3 is not implemented. * However, if we're using the SMC PSCI conduit then QEMU is * effectively acting like EL3 firmware and so the guest at * EL2 should retain the ability to prevent EL1 from being @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo el2_cp_reginfo[] = { .access = PL2_W, .type = ARM_CP_NO_RAW | ARM_CP_EL3_NO_EL2_UNDEF, .writefn = tlbi_aa64_vae2is_write }, #ifndef CONFIG_USER_ONLY - /* Unlike the other EL2-related AT operations, these must + /* + * Unlike the other EL2-related AT operations, these must * UNDEF from EL3 if EL2 is not implemented, which is why we * define them here rather than with the rest of the AT ops. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo el2_cp_reginfo[] = { .access = PL2_W, .accessfn = at_s1e2_access, .type = ARM_CP_NO_RAW | ARM_CP_RAISES_EXC | ARM_CP_EL3_NO_EL2_UNDEF, .writefn = ats_write64 }, - /* The AArch32 ATS1H* operations are CONSTRAINED UNPREDICTABLE + /* + * The AArch32 ATS1H* operations are CONSTRAINED UNPREDICTABLE * if EL2 is not implemented; we choose to UNDEF. Behaviour at EL3 * with SCR.NS == 0 outside Monitor mode is UNPREDICTABLE; we choose * to behave as if SCR.NS was 1. @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo el2_cp_reginfo[] = { .writefn = ats1h_write, .type = ARM_CP_NO_RAW | ARM_CP_RAISES_EXC }, { .name = "CNTHCTL_EL2", .state = ARM_CP_STATE_BOTH, .opc0 = 3, .opc1 = 4, .crn = 14, .crm = 1, .opc2 = 0, - /* ARMv7 requires bit 0 and 1 to reset to 1. ARMv8 defines the + /* + * ARMv7 requires bit 0 and 1 to reset to 1. ARMv8 defines the * reset values as IMPDEF. We choose to reset to 3 to comply with * both ARMv7 and ARMv8. */ @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo el2_sec_cp_reginfo[] = { static CPAccessResult nsacr_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { - /* The NSACR is RW at EL3, and RO for NS EL1 and NS EL2. + /* + * The NSACR is RW at EL3, and RO for NS EL1 and NS EL2. * At Secure EL1 it traps to EL3 or EL2. */ if (arm_current_el(env) == 3) { @@ -XXX,XX +XXX,XX @@ static void define_pmu_regs(ARMCPU *cpu) } } -/* We don't know until after realize whether there's a GICv3 +/* + * We don't know until after realize whether there's a GICv3 * attached, and that is what registers the gicv3 sysregs. * So we have to fill in the GIC fields in ID_PFR/ID_PFR1_EL1/ID_AA64PFR0_EL1 * at runtime. @@ -XXX,XX +XXX,XX @@ static uint64_t id_aa64pfr0_read(CPUARMState *env, const ARMCPRegInfo *ri) } #endif -/* Shared logic between LORID and the rest of the LOR* registers. +/* + * Shared logic between LORID and the rest of the LOR* registers. * Secure state exclusion has already been dealt with. */ static CPAccessResult access_lor_ns(CPUARMState *env, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) define_arm_cp_regs(cpu, cp_reginfo); if (!arm_feature(env, ARM_FEATURE_V8)) { - /* Must go early as it is full of wildcards that may be + /* + * Must go early as it is full of wildcards that may be * overridden by later definitions. */ define_arm_cp_regs(cpu, not_v8_cp_reginfo); @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .access = PL1_R, .type = ARM_CP_CONST, .accessfn = access_aa32_tid3, .resetvalue = cpu->isar.id_pfr0 }, - /* ID_PFR1 is not a plain ARM_CP_CONST because we don't know + /* + * ID_PFR1 is not a plain ARM_CP_CONST because we don't know * the value of the GIC field until after we define these regs. */ { .name = "ID_PFR1", .state = ARM_CP_STATE_BOTH, @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) define_arm_cp_regs(cpu, el3_regs); } - /* The behaviour of NSACR is sufficiently various that we don't + /* + * The behaviour of NSACR is sufficiently various that we don't * try to describe it in a single reginfo: * if EL3 is 64 bit, then trap to EL3 from S EL1, * reads as constant 0xc00 from NS EL1 and NS EL2 @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) if (cpu_isar_feature(aa32_jazelle, cpu)) { define_arm_cp_regs(cpu, jazelle_regs); } - /* Slightly awkwardly, the OMAP and StrongARM cores need all of + /* + * Slightly awkwardly, the OMAP and StrongARM cores need all of * cp15 crn=0 to be writes-ignored, whereas for other cores they should * be read-only (ie write causes UNDEF exception). */ { ARMCPRegInfo id_pre_v8_midr_cp_reginfo[] = { - /* Pre-v8 MIDR space. + /* + * Pre-v8 MIDR space. * Note that the MIDR isn't a simple constant register because * of the TI925 behaviour where writes to another register can * cause the MIDR value to change. @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) if (arm_feature(env, ARM_FEATURE_OMAPCP) || arm_feature(env, ARM_FEATURE_STRONGARM)) { size_t i; - /* Register the blanket "writes ignored" value first to cover the + /* + * Register the blanket "writes ignored" value first to cover the * whole space. Then update the specific ID registers to allow write * access, so that they ignore writes rather than causing them to * UNDEF. @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) .raw_writefn = raw_write, }; if (arm_feature(env, ARM_FEATURE_XSCALE)) { - /* Normally we would always end the TB on an SCTLR write, but Linux + /* + * Normally we would always end the TB on an SCTLR write, but Linux * arch/arm/mach-pxa/sleep.S expects two instructions following * an MMU enable to execute from cache. Imitate this behaviour. */ @@ -XXX,XX +XXX,XX @@ static void add_cpreg_to_hashtable(ARMCPU *cpu, const ARMCPRegInfo *r, void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu, const ARMCPRegInfo *r, void *opaque) { - /* Define implementations of coprocessor registers. + /* + * Define implementations of coprocessor registers. * We store these in a hashtable because typically * there are less than 150 registers in a space which * is 16*16*16*8*8 = 262144 in size. @@ -XXX,XX +XXX,XX @@ void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu, default: g_assert_not_reached(); } - /* The AArch64 pseudocode CheckSystemAccess() specifies that op1 + /* + * The AArch64 pseudocode CheckSystemAccess() specifies that op1 * encodes a minimum access level for the register. We roll this * runtime check into our general permission check code, so check * here that the reginfo's specified permissions are strict enough @@ -XXX,XX +XXX,XX @@ void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu, assert((r->access & ~mask) == 0); } - /* Check that the register definition has enough info to handle + /* + * Check that the register definition has enough info to handle * reads and writes if they are permitted. */ if (!(r->type & (ARM_CP_SPECIAL_MASK | ARM_CP_CONST))) { @@ -XXX,XX +XXX,XX @@ void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu, continue; } if (state == ARM_CP_STATE_AA32) { - /* Under AArch32 CP registers can be common + /* + * Under AArch32 CP registers can be common * (same for secure and non-secure world) or banked. */ char *name; @@ -XXX,XX +XXX,XX @@ void define_one_arm_cp_reg_with_opaque(ARMCPU *cpu, g_assert_not_reached(); } } else { - /* AArch64 registers get mapped to non-secure instance - * of AArch32 */ + /* + * AArch64 registers get mapped to non-secure instance + * of AArch32 + */ add_cpreg_to_hashtable(cpu, r, opaque, state, ARM_CP_SECSTATE_NS, crm, opc1, opc2, r->name); @@ -XXX,XX +XXX,XX @@ void arm_cp_reset_ignore(CPUARMState *env, const ARMCPRegInfo *opaque) static int bad_mode_switch(CPUARMState *env, int mode, CPSRWriteType write_type) { - /* Return true if it is not valid for us to switch to + /* + * Return true if it is not valid for us to switch to * this CPU mode (ie all the UNPREDICTABLE cases in * the ARM ARM CPSRWriteByInstr pseudocode). */ @@ -XXX,XX +XXX,XX @@ static int bad_mode_switch(CPUARMState *env, int mode, CPSRWriteType write_type) case ARM_CPU_MODE_UND: case ARM_CPU_MODE_IRQ: case ARM_CPU_MODE_FIQ: - /* Note that we don't implement the IMPDEF NSACR.RFR which in v7 + /* + * Note that we don't implement the IMPDEF NSACR.RFR which in v7 * allows FIQ mode to be Secure-only. (In v8 this doesn't exist.) */ - /* If HCR.TGE is set then changes from Monitor to NS PL1 via MSR + /* + * If HCR.TGE is set then changes from Monitor to NS PL1 via MSR * and CPS are treated as illegal mode changes. */ if (write_type == CPSRWriteByInstr && @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, env->GE = (val >> 16) & 0xf; } - /* In a V7 implementation that includes the security extensions but does + /* + * In a V7 implementation that includes the security extensions but does * not include Virtualization Extensions the SCR.FW and SCR.AW bits control * whether non-secure software is allowed to change the CPSR_F and CPSR_A * bits respectively. @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, changed_daif = (env->daif ^ val) & mask; if (changed_daif & CPSR_A) { - /* Check to see if we are allowed to change the masking of async + /* + * Check to see if we are allowed to change the masking of async * abort exceptions from a non-secure state. */ if (!(env->cp15.scr_el3 & SCR_AW)) { @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, } if (changed_daif & CPSR_F) { - /* Check to see if we are allowed to change the masking of FIQ + /* + * Check to see if we are allowed to change the masking of FIQ * exceptions from a non-secure state. */ if (!(env->cp15.scr_el3 & SCR_FW)) { @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, mask &= ~CPSR_F; } - /* Check whether non-maskable FIQ (NMFI) support is enabled. + /* + * Check whether non-maskable FIQ (NMFI) support is enabled. * If this bit is set software is not allowed to mask * FIQs, but is allowed to set CPSR_F to 0. */ @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, if (write_type != CPSRWriteRaw && ((env->uncached_cpsr ^ val) & mask & CPSR_M)) { if ((env->uncached_cpsr & CPSR_M) == ARM_CPU_MODE_USR) { - /* Note that we can only get here in USR mode if this is a + /* + * Note that we can only get here in USR mode if this is a * gdb stub write; for this case we follow the architectural * behaviour for guest writes in USR mode of ignoring an attempt * to switch mode. (Those are caught by translate.c for writes @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, */ mask &= ~CPSR_M; } else if (bad_mode_switch(env, val & CPSR_M, write_type)) { - /* Attempt to switch to an invalid mode: this is UNPREDICTABLE in + /* + * Attempt to switch to an invalid mode: this is UNPREDICTABLE in * v7, and has defined behaviour in v8: * + leave CPSR.M untouched * + allow changes to the other CPSR fields @@ -XXX,XX +XXX,XX @@ static void switch_mode(CPUARMState *env, int mode) env->regs[14] = env->banked_r14[r14_bank_number(mode)]; } -/* Physical Interrupt Target EL Lookup Table +/* + * Physical Interrupt Target EL Lookup Table * * [ From ARM ARM section G1.13.4 (Table G1-15) ] * @@ -XXX,XX +XXX,XX @@ uint32_t arm_phys_excp_target_el(CPUState *cs, uint32_t excp_idx, if (arm_feature(env, ARM_FEATURE_EL3)) { rw = ((env->cp15.scr_el3 & SCR_RW) == SCR_RW); } else { - /* Either EL2 is the highest EL (and so the EL2 register width + /* + * Either EL2 is the highest EL (and so the EL2 register width * is given by is64); or there is no EL2 or EL3, in which case * the value of 'rw' does not affect the table lookup anyway. */ @@ -XXX,XX +XXX,XX @@ void aarch64_sync_64_to_32(CPUARMState *env) env->banked_r13[bank_number(ARM_CPU_MODE_UND)] = env->xregs[23]; } - /* Registers x24-x30 are mapped to r8-r14 in FIQ mode. If we are in FIQ + /* + * Registers x24-x30 are mapped to r8-r14 in FIQ mode. If we are in FIQ * mode, then we can copy to r8-r14. Otherwise, we copy to the * FIQ bank for r8-r14. */ @@ -XXX,XX +XXX,XX @@ static void arm_cpu_do_interrupt_aarch32(CPUState *cs) /* High vectors. When enabled, base address cannot be remapped. */ addr += 0xffff0000; } else { - /* ARM v7 architectures provide a vector base address register to remap + /* + * ARM v7 architectures provide a vector base address register to remap * the interrupt vector table. * This register is only followed in non-monitor mode, and is banked. * Note: only bits 31:5 are valid. @@ -XXX,XX +XXX,XX @@ static void arm_cpu_do_interrupt_aarch64(CPUState *cs) aarch64_sve_change_el(env, cur_el, new_el, is_a64(env)); if (cur_el < new_el) { - /* Entry vector offset depends on whether the implemented EL + /* + * Entry vector offset depends on whether the implemented EL * immediately lower than the target level is using AArch32 or AArch64 */ bool is_aa64; @@ -XXX,XX +XXX,XX @@ static void handle_semihosting(CPUState *cs) } #endif -/* Handle a CPU exception for A and R profile CPUs. +/* + * Handle a CPU exception for A and R profile CPUs. * Do any appropriate logging, handle PSCI calls, and then hand off * to the AArch64-entry or AArch32-entry function depending on the * target exception level's register width. @@ -XXX,XX +XXX,XX @@ void arm_cpu_do_interrupt(CPUState *cs) } #endif - /* Hooks may change global state so BQL should be held, also the + /* + * Hooks may change global state so BQL should be held, also the * BQL needs to be held for any modification of * cs->interrupt_request. */ @@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va, }; } -/* Note that signed overflow is undefined in C. The following routines are - careful to use unsigned types where modulo arithmetic is required. - Failure to do so _will_ break on newer gcc. */ +/* + * Note that signed overflow is undefined in C. The following routines are + * careful to use unsigned types where modulo arithmetic is required. + * Failure to do so _will_ break on newer gcc. + */ /* Signed saturating arithmetic. */ @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(sel_flags)(uint32_t flags, uint32_t a, uint32_t b) return (a & mask) | (b & ~mask); } -/* CRC helpers. +/* + * CRC helpers. * The upper bytes of val (above the number specified by 'bytes') must have * been zeroed out by the caller. */ @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(crc32c)(uint32_t acc, uint32_t val, uint32_t bytes) return crc32c(acc, buf, bytes) ^ 0xffffffff; } -/* Return the exception level to which FP-disabled exceptions should +/* + * Return the exception level to which FP-disabled exceptions should * be taken, or 0 if FP is enabled. */ int fp_exception_el(CPUARMState *env, int cur_el) @@ -XXX,XX +XXX,XX @@ int fp_exception_el(CPUARMState *env, int cur_el) #ifndef CONFIG_USER_ONLY uint64_t hcr_el2; - /* CPACR and the CPTR registers don't exist before v6, so FP is + /* + * CPACR and the CPTR registers don't exist before v6, so FP is * always accessible */ if (!arm_feature(env, ARM_FEATURE_V6)) { @@ -XXX,XX +XXX,XX @@ int fp_exception_el(CPUARMState *env, int cur_el) hcr_el2 = arm_hcr_el2_eff(env); - /* The CPACR controls traps to EL1, or PL1 if we're 32 bit: + /* + * The CPACR controls traps to EL1, or PL1 if we're 32 bit: * 0, 2 : trap EL0 and EL1/PL1 accesses * 1 : trap only EL0 accesses * 3 : trap no accesses -- 2.25.1 From: Fabiano Rosas <farosas@suse.de> Fix the following: ERROR: spaces required around that '|' (ctx:VxV) ERROR: space required before the open parenthesis '(' ERROR: spaces required around that '+' (ctx:VxB) ERROR: space prohibited between function name and open parenthesis '(' (the last two still have some occurrences in macros which I left behind because it might impact readability) Signed-off-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Message-id: 20221213190537.511-3-farosas@suse.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 42 +++++++++++++++++++++--------------------- 1 file changed, 21 insertions(+), 21 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ static void add_cpreg_to_list(gpointer key, gpointer opaque) uint32_t regidx = (uintptr_t)key; const ARMCPRegInfo *ri = get_arm_cp_reginfo(cpu->cp_regs, regidx); - if (!(ri->type & (ARM_CP_NO_RAW|ARM_CP_ALIAS))) { + if (!(ri->type & (ARM_CP_NO_RAW | ARM_CP_ALIAS))) { cpu->cpreg_indexes[cpu->cpreg_array_len] = cpreg_to_kvm_id(regidx); /* The value array need not be initialized at this point */ cpu->cpreg_array_len++; @@ -XXX,XX +XXX,XX @@ static void count_cpreg(gpointer key, gpointer opaque) ri = g_hash_table_lookup(cpu->cp_regs, key); - if (!(ri->type & (ARM_CP_NO_RAW|ARM_CP_ALIAS))) { + if (!(ri->type & (ARM_CP_NO_RAW | ARM_CP_ALIAS))) { cpu->cpreg_array_len++; } } @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v6k_cp_reginfo[] = { .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, + .access = PL0_R | PL1_W, .fieldoffset = offsetof(CPUARMState, cp15.tpidrro_el[0]), .resetvalue = 0}, { .name = "TPIDRURO", .cp = 15, .crn = 13, .crm = 0, .opc1 = 0, .opc2 = 3, - .access = PL0_R|PL1_W, + .access = PL0_R | PL1_W, .bank_fieldoffsets = { offsetoflow32(CPUARMState, cp15.tpidruro_s), offsetoflow32(CPUARMState, cp15.tpidruro_ns) }, .resetfn = arm_cp_reset_ignore }, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo cache_block_ops_cp_reginfo[] = { .resetvalue = 0 }, /* The cache ops themselves: these all NOP for QEMU */ { .name = "IICR", .cp = 15, .crm = 5, .opc1 = 0, - .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL1_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, { .name = "IDCR", .cp = 15, .crm = 6, .opc1 = 0, - .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL1_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, { .name = "CDCR", .cp = 15, .crm = 12, .opc1 = 0, - .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL0_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, { .name = "PIR", .cp = 15, .crm = 12, .opc1 = 1, - .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL0_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, { .name = "PDR", .cp = 15, .crm = 12, .opc1 = 2, - .access = PL0_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL0_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, { .name = "CIDCR", .cp = 15, .crm = 14, .opc1 = 0, - .access = PL1_W, .type = ARM_CP_NOP|ARM_CP_64BIT }, + .access = PL1_W, .type = ARM_CP_NOP | ARM_CP_64BIT }, }; static const ARMCPRegInfo cache_test_clean_cp_reginfo[] = { @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) ARMCPRegInfo cbar = { .name = "CBAR", .cp = 15, .crn = 15, .crm = 0, .opc1 = 4, .opc2 = 0, - .access = PL1_R|PL3_W, .resetvalue = cpu->reset_cbar, + .access = PL1_R | PL3_W, .resetvalue = cpu->reset_cbar, .fieldoffset = offsetof(CPUARMState, cp15.c15_config_base_address) }; @@ -XXX,XX +XXX,XX @@ static void switch_mode(CPUARMState *env, int mode) return; if (old_mode == ARM_CPU_MODE_FIQ) { - memcpy (env->fiq_regs, env->regs + 8, 5 * sizeof(uint32_t)); - memcpy (env->regs + 8, env->usr_regs, 5 * sizeof(uint32_t)); + memcpy(env->fiq_regs, env->regs + 8, 5 * sizeof(uint32_t)); + memcpy(env->regs + 8, env->usr_regs, 5 * sizeof(uint32_t)); } else if (mode == ARM_CPU_MODE_FIQ) { - memcpy (env->usr_regs, env->regs + 8, 5 * sizeof(uint32_t)); - memcpy (env->regs + 8, env->fiq_regs, 5 * sizeof(uint32_t)); + memcpy(env->usr_regs, env->regs + 8, 5 * sizeof(uint32_t)); + memcpy(env->regs + 8, env->fiq_regs, 5 * sizeof(uint32_t)); } i = bank_number(old_mode); @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 16); \ if (sum >= 0) \ ge |= 3 << (n * 2); \ - } while(0) + } while (0) #define SARITH8(a, b, n, op) do { \ int32_t sum; \ @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 8); \ if (sum >= 0) \ ge |= 1 << n; \ - } while(0) + } while (0) #define ADD16(a, b, n) SARITH16(a, b, n, +) @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 16); \ if ((sum >> 16) == 1) \ ge |= 3 << (n * 2); \ - } while(0) + } while (0) #define ADD8(a, b, n) do { \ uint32_t sum; \ @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 8); \ if ((sum >> 8) == 1) \ ge |= 1 << n; \ - } while(0) + } while (0) #define SUB16(a, b, n) do { \ uint32_t sum; \ @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 16); \ if ((sum >> 16) == 0) \ ge |= 3 << (n * 2); \ - } while(0) + } while (0) #define SUB8(a, b, n) do { \ uint32_t sum; \ @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) RESULT(sum, n, 8); \ if ((sum >> 8) == 0) \ ge |= 1 << n; \ - } while(0) + } while (0) #define PFX u #define ARITH_GE -- 2.25.1 From: Fabiano Rosas <farosas@suse.de> Fix this: ERROR: braces {} are necessary for all arms of this statement Signed-off-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Message-id: 20221213190537.511-4-farosas@suse.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 67 ++++++++++++++++++++++++++++----------------- 1 file changed, 42 insertions(+), 25 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void cpsr_write(CPUARMState *env, uint32_t val, uint32_t mask, env->CF = (val >> 29) & 1; env->VF = (val << 3) & 0x80000000; } - if (mask & CPSR_Q) + if (mask & CPSR_Q) { env->QF = ((val & CPSR_Q) != 0); - if (mask & CPSR_T) + } + if (mask & CPSR_T) { env->thumb = ((val & CPSR_T) != 0); + } if (mask & CPSR_IT_0_1) { env->condexec_bits &= ~3; env->condexec_bits |= (val >> 25) & 3; @@ -XXX,XX +XXX,XX @@ static void switch_mode(CPUARMState *env, int mode) int i; old_mode = env->uncached_cpsr & CPSR_M; - if (mode == old_mode) + if (mode == old_mode) { return; + } if (old_mode == ARM_CPU_MODE_FIQ) { memcpy(env->fiq_regs, env->regs + 8, 5 * sizeof(uint32_t)); @@ -XXX,XX +XXX,XX @@ static void arm_cpu_do_interrupt_aarch32(CPUState *cs) new_mode = ARM_CPU_MODE_UND; addr = 0x04; mask = CPSR_I; - if (env->thumb) + if (env->thumb) { offset = 2; - else + } else { offset = 4; + } break; case EXCP_SWI: new_mode = ARM_CPU_MODE_SVC; @@ -XXX,XX +XXX,XX @@ static inline uint16_t add16_sat(uint16_t a, uint16_t b) res = a + b; if (((res ^ a) & 0x8000) && !((a ^ b) & 0x8000)) { - if (a & 0x8000) + if (a & 0x8000) { res = 0x8000; - else + } else { res = 0x7fff; + } } return res; } @@ -XXX,XX +XXX,XX @@ static inline uint8_t add8_sat(uint8_t a, uint8_t b) res = a + b; if (((res ^ a) & 0x80) && !((a ^ b) & 0x80)) { - if (a & 0x80) + if (a & 0x80) { res = 0x80; - else + } else { res = 0x7f; + } } return res; } @@ -XXX,XX +XXX,XX @@ static inline uint16_t sub16_sat(uint16_t a, uint16_t b) res = a - b; if (((res ^ a) & 0x8000) && ((a ^ b) & 0x8000)) { - if (a & 0x8000) + if (a & 0x8000) { res = 0x8000; - else + } else { res = 0x7fff; + } } return res; } @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_sat(uint8_t a, uint8_t b) res = a - b; if (((res ^ a) & 0x80) && ((a ^ b) & 0x80)) { - if (a & 0x80) + if (a & 0x80) { res = 0x80; - else + } else { res = 0x7f; + } } return res; } @@ -XXX,XX +XXX,XX @@ static inline uint16_t add16_usat(uint16_t a, uint16_t b) { uint16_t res; res = a + b; - if (res < a) + if (res < a) { res = 0xffff; + } return res; } static inline uint16_t sub16_usat(uint16_t a, uint16_t b) { - if (a > b) + if (a > b) { return a - b; - else + } else { return 0; + } } static inline uint8_t add8_usat(uint8_t a, uint8_t b) { uint8_t res; res = a + b; - if (res < a) + if (res < a) { res = 0xff; + } return res; } static inline uint8_t sub8_usat(uint8_t a, uint8_t b) { - if (a > b) + if (a > b) { return a - b; - else + } else { return 0; + } } #define ADD16(a, b, n) RESULT(add16_usat(a, b), n, 16); @@ -XXX,XX +XXX,XX @@ static inline uint8_t sub8_usat(uint8_t a, uint8_t b) static inline uint8_t do_usad(uint8_t a, uint8_t b) { - if (a > b) + if (a > b) { return a - b; - else + } else { return b - a; + } } /* Unsigned sum of absolute byte differences. */ @@ -XXX,XX +XXX,XX @@ uint32_t HELPER(sel_flags)(uint32_t flags, uint32_t a, uint32_t b) uint32_t mask; mask = 0; - if (flags & 1) + if (flags & 1) { mask |= 0xff; - if (flags & 2) + } + if (flags & 2) { mask |= 0xff00; - if (flags & 4) + } + if (flags & 4) { mask |= 0xff0000; - if (flags & 8) + } + if (flags & 8) { mask |= 0xff000000; + } return (a & mask) | (b & ~mask); } -- 2.25.1 From: Fabiano Rosas <farosas@suse.de> Signed-off-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Message-id: 20221213190537.511-5-farosas@suse.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/m_helper.c | 16 ---------------- 1 file changed, 16 deletions(-) diff --git a/target/arm/m_helper.c b/target/arm/m_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/m_helper.c +++ b/target/arm/m_helper.c @@ -XXX,XX +XXX,XX @@ */ #include "qemu/osdep.h" -#include "qemu/units.h" -#include "target/arm/idau.h" -#include "trace.h" #include "cpu.h" #include "internals.h" -#include "exec/gdbstub.h" #include "exec/helper-proto.h" -#include "qemu/host-utils.h" #include "qemu/main-loop.h" #include "qemu/bitops.h" -#include "qemu/crc32c.h" -#include "qemu/qemu-print.h" #include "qemu/log.h" #include "exec/exec-all.h" -#include <zlib.h> /* For crc32 */ -#include "semihosting/semihost.h" -#include "sysemu/cpus.h" -#include "sysemu/kvm.h" -#include "qemu/range.h" -#include "qapi/qapi-commands-machine-target.h" -#include "qapi/error.h" -#include "qemu/guest-random.h" #ifdef CONFIG_TCG -#include "arm_ldst.h" #include "exec/cpu_ldst.h" #include "semihosting/common-semi.h" #endif -- 2.25.1 From: Fabiano Rosas <farosas@suse.de> Signed-off-by: Fabiano Rosas <farosas@suse.de> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Message-id: 20221213190537.511-6-farosas@suse.de Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ */ #include "qemu/osdep.h" -#include "qemu/units.h" #include "qemu/log.h" #include "trace.h" #include "cpu.h" #include "internals.h" #include "exec/helper-proto.h" -#include "qemu/host-utils.h" #include "qemu/main-loop.h" #include "qemu/timer.h" #include "qemu/bitops.h" @@ -XXX,XX +XXX,XX @@ #include "exec/exec-all.h" #include <zlib.h> /* For crc32 */ #include "hw/irq.h" -#include "semihosting/semihost.h" -#include "sysemu/cpus.h" #include "sysemu/cpu-timers.h" #include "sysemu/kvm.h" -#include "qemu/range.h" #include "qapi/qapi-commands-machine-target.h" #include "qapi/error.h" #include "qemu/guest-random.h" #ifdef CONFIG_TCG -#include "arm_ldst.h" -#include "exec/cpu_ldst.h" #include "semihosting/common-semi.h" #endif #include "cpregs.h" -- 2.25.1 From: Claudio Fontana <cfontana@suse.de> Remove some unused headers. Signed-off-by: Claudio Fontana <cfontana@suse.de> Acked-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Claudio Fontana <cfontana@suse.de> Reviewed-by: Cornelia Huck <cohuck@redhat.com> Signed-off-by: Fabiano Rosas <farosas@suse.de> Message-id: 20221213190537.511-7-farosas@suse.de [added back some includes that are still needed at this point] Signed-off-by: Fabiano Rosas <farosas@suse.de> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.c | 1 - target/arm/cpu64.c | 6 ------ 2 files changed, 7 deletions(-) diff --git a/target/arm/cpu.c b/target/arm/cpu.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu.c +++ b/target/arm/cpu.c @@ -XXX,XX +XXX,XX @@ #include "target/arm/idau.h" #include "qemu/module.h" #include "qapi/error.h" -#include "qapi/visitor.h" #include "cpu.h" #ifdef CONFIG_TCG #include "hw/core/tcg-cpu-ops.h" diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu64.c +++ b/target/arm/cpu64.c @@ -XXX,XX +XXX,XX @@ #include "qemu/osdep.h" #include "qapi/error.h" #include "cpu.h" -#ifdef CONFIG_TCG -#include "hw/core/tcg-cpu-ops.h" -#endif /* CONFIG_TCG */ #include "qemu/module.h" -#if !defined(CONFIG_USER_ONLY) -#include "hw/loader.h" -#endif #include "sysemu/kvm.h" #include "sysemu/hvf.h" #include "kvm_arm.h" -- 2.25.1 From: Philippe Mathieu-Daudé <philmd@linaro.org> The pointed MouseTransformInfo structure is accessed read-only. Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20221220142520.24094-2-philmd@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/input/tsc2xxx.h | 4 ++-- hw/input/tsc2005.c | 2 +- hw/input/tsc210x.c | 3 +-- 3 files changed, 4 insertions(+), 5 deletions(-) diff --git a/include/hw/input/tsc2xxx.h b/include/hw/input/tsc2xxx.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/input/tsc2xxx.h +++ b/include/hw/input/tsc2xxx.h @@ -XXX,XX +XXX,XX @@ uWireSlave *tsc2102_init(qemu_irq pint); uWireSlave *tsc2301_init(qemu_irq penirq, qemu_irq kbirq, qemu_irq dav); I2SCodec *tsc210x_codec(uWireSlave *chip); uint32_t tsc210x_txrx(void *opaque, uint32_t value, int len); -void tsc210x_set_transform(uWireSlave *chip, MouseTransformInfo *info); +void tsc210x_set_transform(uWireSlave *chip, const MouseTransformInfo *info); void tsc210x_key_event(uWireSlave *chip, int key, int down); /* tsc2005.c */ void *tsc2005_init(qemu_irq pintdav); uint32_t tsc2005_txrx(void *opaque, uint32_t value, int len); -void tsc2005_set_transform(void *opaque, MouseTransformInfo *info); +void tsc2005_set_transform(void *opaque, const MouseTransformInfo *info); #endif diff --git a/hw/input/tsc2005.c b/hw/input/tsc2005.c index XXXXXXX..XXXXXXX 100644 --- a/hw/input/tsc2005.c +++ b/hw/input/tsc2005.c @@ -XXX,XX +XXX,XX @@ void *tsc2005_init(qemu_irq pintdav) * from the touchscreen. Assuming 12-bit precision was used during * tslib calibration. */ -void tsc2005_set_transform(void *opaque, MouseTransformInfo *info) +void tsc2005_set_transform(void *opaque, const MouseTransformInfo *info) { TSC2005State *s = (TSC2005State *) opaque; diff --git a/hw/input/tsc210x.c b/hw/input/tsc210x.c index XXXXXXX..XXXXXXX 100644 --- a/hw/input/tsc210x.c +++ b/hw/input/tsc210x.c @@ -XXX,XX +XXX,XX @@ I2SCodec *tsc210x_codec(uWireSlave *chip) * from the touchscreen. Assuming 12-bit precision was used during * tslib calibration. */ -void tsc210x_set_transform(uWireSlave *chip, - MouseTransformInfo *info) +void tsc210x_set_transform(uWireSlave *chip, const MouseTransformInfo *info) { TSC210xState *s = (TSC210xState *) chip->opaque; #if 0 -- 2.25.1 From: Philippe Mathieu-Daudé <philmd@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20221220142520.24094-3-philmd@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/nseries.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/hw/arm/nseries.c b/hw/arm/nseries.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/nseries.c +++ b/hw/arm/nseries.c @@ -XXX,XX +XXX,XX @@ static void n8x0_i2c_setup(struct n800_s *s) } /* Touchscreen and keypad controller */ -static MouseTransformInfo n800_pointercal = { +static const MouseTransformInfo n800_pointercal = { .x = 800, .y = 480, .a = { 14560, -68, -3455208, -39, -9621, 35152972, 65536 }, }; -static MouseTransformInfo n810_pointercal = { +static const MouseTransformInfo n810_pointercal = { .x = 800, .y = 480, .a = { 15041, 148, -4731056, 171, -10238, 35933380, 65536 }, @@ -XXX,XX +XXX,XX @@ static void n810_key_event(void *opaque, int keycode) #define M 0 -static int n810_keys[0x80] = { +static const int n810_keys[0x80] = { [0x01] = 16, /* Q */ [0x02] = 37, /* K */ [0x03] = 24, /* O */ @@ -XXX,XX +XXX,XX @@ static void n8x0_usb_setup(struct n800_s *s) /* Setup done before the main bootloader starts by some early setup code * - used when we want to run the main bootloader in emulation. This * isn't documented. */ -static uint32_t n800_pinout[104] = { +static const uint32_t n800_pinout[104] = { 0x080f00d8, 0x00d40808, 0x03080808, 0x080800d0, 0x00dc0808, 0x0b0f0f00, 0x080800b4, 0x00c00808, 0x08080808, 0x180800c4, 0x00b80000, 0x08080808, @@ -XXX,XX +XXX,XX @@ static void n8x0_boot_init(void *opaque) #define OMAP_TAG_CBUS 0x4e03 #define OMAP_TAG_EM_ASIC_BB5 0x4e04 -static struct omap_gpiosw_info_s { +static const struct omap_gpiosw_info_s { const char *name; int line; int type; @@ -XXX,XX +XXX,XX @@ static struct omap_gpiosw_info_s { { NULL } }; -static struct omap_partition_info_s { +static const struct omap_partition_info_s { uint32_t offset; uint32_t size; int mask; @@ -XXX,XX +XXX,XX @@ static struct omap_partition_info_s { { 0, 0, 0, NULL } }; -static uint8_t n8x0_bd_addr[6] = { N8X0_BD_ADDR }; +static const uint8_t n8x0_bd_addr[6] = { N8X0_BD_ADDR }; static int n8x0_atag_setup(void *p, int model) { uint8_t *b; uint16_t *w; uint32_t *l; - struct omap_gpiosw_info_s *gpiosw; - struct omap_partition_info_s *partition; + const struct omap_gpiosw_info_s *gpiosw; + const struct omap_partition_info_s *partition; const char *tag; w = p; -- 2.25.1 From: Philippe Mathieu-Daudé <philmd@linaro.org> Silent when compiling with -Wextra: ../hw/arm/nseries.c:1081:12: warning: missing field 'line' initializer [-Wmissing-field-initializers] { NULL } ^ Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-id: 20221220142520.24094-4-philmd@linaro.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/nseries.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/hw/arm/nseries.c b/hw/arm/nseries.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/nseries.c +++ b/hw/arm/nseries.c @@ -XXX,XX +XXX,XX @@ static const struct omap_gpiosw_info_s { "headphone", N8X0_HEADPHONE_GPIO, OMAP_GPIOSW_TYPE_CONNECTION | OMAP_GPIOSW_INVERTED, }, - { NULL } + { /* end of list */ } }, n810_gpiosw_info[] = { { "gps_reset", N810_GPS_RESET_GPIO, @@ -XXX,XX +XXX,XX @@ static const struct omap_gpiosw_info_s { "slide", N810_SLIDE_GPIO, OMAP_GPIOSW_TYPE_COVER | OMAP_GPIOSW_INVERTED, }, - { NULL } + { /* end of list */ } }; static const struct omap_partition_info_s { @@ -XXX,XX +XXX,XX @@ static const struct omap_partition_info_s { { 0x00080000, 0x00200000, 0x0, "kernel" }, { 0x00280000, 0x00200000, 0x3, "initfs" }, { 0x00480000, 0x0fb80000, 0x3, "rootfs" }, - - { 0, 0, 0, NULL } + { /* end of list */ } }, n810_part_info[] = { { 0x00000000, 0x00020000, 0x3, "bootloader" }, { 0x00020000, 0x00060000, 0x0, "config" }, { 0x00080000, 0x00220000, 0x0, "kernel" }, { 0x002a0000, 0x00400000, 0x0, "initfs" }, { 0x006a0000, 0x0f960000, 0x0, "rootfs" }, - - { 0, 0, 0, NULL } + { /* end of list */ } }; static const uint8_t n8x0_bd_addr[6] = { N8X0_BD_ADDR }; -- 2.25.1 From: Zhuojia Shen <chaosdefinition@hotmail.com> In CPUID registers exposed to userspace, some registers were missing and some fields were not exposed. This patch aligns exposed ID registers and their fields with what the upstream kernel currently exposes. Specifically, the following new ID registers/fields are exposed to userspace: ID_AA64PFR1_EL1.BT: bits 3-0 ID_AA64PFR1_EL1.MTE: bits 11-8 ID_AA64PFR1_EL1.SME: bits 27-24 ID_AA64ZFR0_EL1.SVEver: bits 3-0 ID_AA64ZFR0_EL1.AES: bits 7-4 ID_AA64ZFR0_EL1.BitPerm: bits 19-16 ID_AA64ZFR0_EL1.BF16: bits 23-20 ID_AA64ZFR0_EL1.SHA3: bits 35-32 ID_AA64ZFR0_EL1.SM4: bits 43-40 ID_AA64ZFR0_EL1.I8MM: bits 47-44 ID_AA64ZFR0_EL1.F32MM: bits 55-52 ID_AA64ZFR0_EL1.F64MM: bits 59-56 ID_AA64SMFR0_EL1.F32F32: bit 32 ID_AA64SMFR0_EL1.B16F32: bit 34 ID_AA64SMFR0_EL1.F16F32: bit 35 ID_AA64SMFR0_EL1.I8I32: bits 39-36 ID_AA64SMFR0_EL1.F64F64: bit 48 ID_AA64SMFR0_EL1.I16I64: bits 55-52 ID_AA64SMFR0_EL1.FA64: bit 63 ID_AA64MMFR0_EL1.ECV: bits 63-60 ID_AA64MMFR1_EL1.AFP: bits 47-44 ID_AA64MMFR2_EL1.AT: bits 35-32 ID_AA64ISAR0_EL1.RNDR: bits 63-60 ID_AA64ISAR1_EL1.FRINTTS: bits 35-32 ID_AA64ISAR1_EL1.BF16: bits 47-44 ID_AA64ISAR1_EL1.DGH: bits 51-48 ID_AA64ISAR1_EL1.I8MM: bits 55-52 ID_AA64ISAR2_EL1.WFxT: bits 3-0 ID_AA64ISAR2_EL1.RPRES: bits 7-4 ID_AA64ISAR2_EL1.GPA3: bits 11-8 ID_AA64ISAR2_EL1.APA3: bits 15-12 The code is also refactored to use symbolic names for ID register fields for better readability and maintainability. The test case in tests/tcg/aarch64/sysregs.c is also updated to match the intended behavior. Signed-off-by: Zhuojia Shen <chaosdefinition@hotmail.com> Message-id: DS7PR12MB6309FB585E10772928F14271ACE79@DS7PR12MB6309.namprd12.prod.outlook.com Reviewed-by: Peter Maydell <peter.maydell@linaro.org> [PMM: use Sn_n_Cn_Cn_n syntax to work with older assemblers that don't recognize id_aa64isar2_el1 and id_aa64mmfr2_el1] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/helper.c | 96 +++++++++++++++++++++++++------ tests/tcg/aarch64/sysregs.c | 24 ++++++-- tests/tcg/aarch64/Makefile.target | 7 ++- 3 files changed, 103 insertions(+), 24 deletions(-) diff --git a/target/arm/helper.c b/target/arm/helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/helper.c +++ b/target/arm/helper.c @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) #ifdef CONFIG_USER_ONLY static const ARMCPRegUserSpaceInfo v8_user_idregs[] = { { .name = "ID_AA64PFR0_EL1", - .exported_bits = 0x000f000f00ff0000, - .fixed_bits = 0x0000000000000011 }, + .exported_bits = R_ID_AA64PFR0_FP_MASK | + R_ID_AA64PFR0_ADVSIMD_MASK | + R_ID_AA64PFR0_SVE_MASK | + R_ID_AA64PFR0_DIT_MASK, + .fixed_bits = (0x1u << R_ID_AA64PFR0_EL0_SHIFT) | + (0x1u << R_ID_AA64PFR0_EL1_SHIFT) }, { .name = "ID_AA64PFR1_EL1", - .exported_bits = 0x00000000000000f0 }, + .exported_bits = R_ID_AA64PFR1_BT_MASK | + R_ID_AA64PFR1_SSBS_MASK | + R_ID_AA64PFR1_MTE_MASK | + R_ID_AA64PFR1_SME_MASK }, { .name = "ID_AA64PFR*_EL1_RESERVED", - .is_glob = true }, - { .name = "ID_AA64ZFR0_EL1" }, + .is_glob = true }, + { .name = "ID_AA64ZFR0_EL1", + .exported_bits = R_ID_AA64ZFR0_SVEVER_MASK | + R_ID_AA64ZFR0_AES_MASK | + R_ID_AA64ZFR0_BITPERM_MASK | + R_ID_AA64ZFR0_BFLOAT16_MASK | + R_ID_AA64ZFR0_SHA3_MASK | + R_ID_AA64ZFR0_SM4_MASK | + R_ID_AA64ZFR0_I8MM_MASK | + R_ID_AA64ZFR0_F32MM_MASK | + R_ID_AA64ZFR0_F64MM_MASK }, + { .name = "ID_AA64SMFR0_EL1", + .exported_bits = R_ID_AA64SMFR0_F32F32_MASK | + R_ID_AA64SMFR0_B16F32_MASK | + R_ID_AA64SMFR0_F16F32_MASK | + R_ID_AA64SMFR0_I8I32_MASK | + R_ID_AA64SMFR0_F64F64_MASK | + R_ID_AA64SMFR0_I16I64_MASK | + R_ID_AA64SMFR0_FA64_MASK }, { .name = "ID_AA64MMFR0_EL1", - .fixed_bits = 0x00000000ff000000 }, - { .name = "ID_AA64MMFR1_EL1" }, + .exported_bits = R_ID_AA64MMFR0_ECV_MASK, + .fixed_bits = (0xfu << R_ID_AA64MMFR0_TGRAN64_SHIFT) | + (0xfu << R_ID_AA64MMFR0_TGRAN4_SHIFT) }, + { .name = "ID_AA64MMFR1_EL1", + .exported_bits = R_ID_AA64MMFR1_AFP_MASK }, + { .name = "ID_AA64MMFR2_EL1", + .exported_bits = R_ID_AA64MMFR2_AT_MASK }, { .name = "ID_AA64MMFR*_EL1_RESERVED", - .is_glob = true }, + .is_glob = true }, { .name = "ID_AA64DFR0_EL1", - .fixed_bits = 0x0000000000000006 }, - { .name = "ID_AA64DFR1_EL1" }, + .fixed_bits = (0x6u << R_ID_AA64DFR0_DEBUGVER_SHIFT) }, + { .name = "ID_AA64DFR1_EL1" }, { .name = "ID_AA64DFR*_EL1_RESERVED", - .is_glob = true }, + .is_glob = true }, { .name = "ID_AA64AFR*", - .is_glob = true }, + .is_glob = true }, { .name = "ID_AA64ISAR0_EL1", - .exported_bits = 0x00fffffff0fffff0 }, + .exported_bits = R_ID_AA64ISAR0_AES_MASK | + R_ID_AA64ISAR0_SHA1_MASK | + R_ID_AA64ISAR0_SHA2_MASK | + R_ID_AA64ISAR0_CRC32_MASK | + R_ID_AA64ISAR0_ATOMIC_MASK | + R_ID_AA64ISAR0_RDM_MASK | + R_ID_AA64ISAR0_SHA3_MASK | + R_ID_AA64ISAR0_SM3_MASK | + R_ID_AA64ISAR0_SM4_MASK | + R_ID_AA64ISAR0_DP_MASK | + R_ID_AA64ISAR0_FHM_MASK | + R_ID_AA64ISAR0_TS_MASK | + R_ID_AA64ISAR0_RNDR_MASK }, { .name = "ID_AA64ISAR1_EL1", - .exported_bits = 0x000000f0ffffffff }, + .exported_bits = R_ID_AA64ISAR1_DPB_MASK | + R_ID_AA64ISAR1_APA_MASK | + R_ID_AA64ISAR1_API_MASK | + R_ID_AA64ISAR1_JSCVT_MASK | + R_ID_AA64ISAR1_FCMA_MASK | + R_ID_AA64ISAR1_LRCPC_MASK | + R_ID_AA64ISAR1_GPA_MASK | + R_ID_AA64ISAR1_GPI_MASK | + R_ID_AA64ISAR1_FRINTTS_MASK | + R_ID_AA64ISAR1_SB_MASK | + R_ID_AA64ISAR1_BF16_MASK | + R_ID_AA64ISAR1_DGH_MASK | + R_ID_AA64ISAR1_I8MM_MASK }, + { .name = "ID_AA64ISAR2_EL1", + .exported_bits = R_ID_AA64ISAR2_WFXT_MASK | + R_ID_AA64ISAR2_RPRES_MASK | + R_ID_AA64ISAR2_GPA3_MASK | + R_ID_AA64ISAR2_APA3_MASK }, { .name = "ID_AA64ISAR*_EL1_RESERVED", - .is_glob = true }, + .is_glob = true }, }; modify_arm_cp_regs(v8_idregs, v8_user_idregs); #endif @@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu) #ifdef CONFIG_USER_ONLY static const ARMCPRegUserSpaceInfo id_v8_user_midr_cp_reginfo[] = { { .name = "MIDR_EL1", - .exported_bits = 0x00000000ffffffff }, - { .name = "REVIDR_EL1" }, + .exported_bits = R_MIDR_EL1_REVISION_MASK | + R_MIDR_EL1_PARTNUM_MASK | + R_MIDR_EL1_ARCHITECTURE_MASK | + R_MIDR_EL1_VARIANT_MASK | + R_MIDR_EL1_IMPLEMENTER_MASK }, + { .name = "REVIDR_EL1" }, }; modify_arm_cp_regs(id_v8_midr_cp_reginfo, id_v8_user_midr_cp_reginfo); #endif diff --git a/tests/tcg/aarch64/sysregs.c b/tests/tcg/aarch64/sysregs.c index XXXXXXX..XXXXXXX 100644 --- a/tests/tcg/aarch64/sysregs.c +++ b/tests/tcg/aarch64/sysregs.c @@ -XXX,XX +XXX,XX @@ #define HWCAP_CPUID (1 << 11) #endif +/* + * Older assemblers don't recognize newer system register names, + * but we can still access them by the Sn_n_Cn_Cn_n syntax. + */ +#define SYS_ID_AA64ISAR2_EL1 S3_0_C0_C6_2 +#define SYS_ID_AA64MMFR2_EL1 S3_0_C0_C7_2 + int failed_bit_count; /* Read and print system register `id' value */ @@ -XXX,XX +XXX,XX @@ int main(void) * minimum valid fields - for the purposes of this check allowed * to have non-zero values. */ - get_cpu_reg_check_mask(id_aa64isar0_el1, _m(00ff,ffff,f0ff,fff0)); - get_cpu_reg_check_mask(id_aa64isar1_el1, _m(0000,00f0,ffff,ffff)); + get_cpu_reg_check_mask(id_aa64isar0_el1, _m(f0ff,ffff,f0ff,fff0)); + get_cpu_reg_check_mask(id_aa64isar1_el1, _m(00ff,f0ff,ffff,ffff)); + get_cpu_reg_check_mask(SYS_ID_AA64ISAR2_EL1, _m(0000,0000,0000,ffff)); /* TGran4 & TGran64 as pegged to -1 */ - get_cpu_reg_check_mask(id_aa64mmfr0_el1, _m(0000,0000,ff00,0000)); - get_cpu_reg_check_zero(id_aa64mmfr1_el1); + get_cpu_reg_check_mask(id_aa64mmfr0_el1, _m(f000,0000,ff00,0000)); + get_cpu_reg_check_mask(id_aa64mmfr1_el1, _m(0000,f000,0000,0000)); + get_cpu_reg_check_mask(SYS_ID_AA64MMFR2_EL1, _m(0000,000f,0000,0000)); /* EL1/EL0 reported as AA64 only */ get_cpu_reg_check_mask(id_aa64pfr0_el1, _m(000f,000f,00ff,0011)); - get_cpu_reg_check_mask(id_aa64pfr1_el1, _m(0000,0000,0000,00f0)); + get_cpu_reg_check_mask(id_aa64pfr1_el1, _m(0000,0000,0f00,0fff)); /* all hidden, DebugVer fixed to 0x6 (ARMv8 debug architecture) */ get_cpu_reg_check_mask(id_aa64dfr0_el1, _m(0000,0000,0000,0006)); get_cpu_reg_check_zero(id_aa64dfr1_el1); - get_cpu_reg_check_zero(id_aa64zfr0_el1); + get_cpu_reg_check_mask(id_aa64zfr0_el1, _m(0ff0,ff0f,00ff,00ff)); +#ifdef HAS_ARMV9_SME + get_cpu_reg_check_mask(id_aa64smfr0_el1, _m(80f1,00fd,0000,0000)); +#endif get_cpu_reg_check_zero(id_aa64afr0_el1); get_cpu_reg_check_zero(id_aa64afr1_el1); diff --git a/tests/tcg/aarch64/Makefile.target b/tests/tcg/aarch64/Makefile.target index XXXXXXX..XXXXXXX 100644 --- a/tests/tcg/aarch64/Makefile.target +++ b/tests/tcg/aarch64/Makefile.target @@ -XXX,XX +XXX,XX @@ config-cc.mak: Makefile $(call cc-option,-march=armv8.1-a+sve2, CROSS_CC_HAS_SVE2); \ $(call cc-option,-march=armv8.3-a, CROSS_CC_HAS_ARMV8_3); \ $(call cc-option,-mbranch-protection=standard, CROSS_CC_HAS_ARMV8_BTI); \ - $(call cc-option,-march=armv8.5-a+memtag, CROSS_CC_HAS_ARMV8_MTE)) 3> config-cc.mak + $(call cc-option,-march=armv8.5-a+memtag, CROSS_CC_HAS_ARMV8_MTE); \ + $(call cc-option,-march=armv9-a+sme, CROSS_CC_HAS_ARMV9_SME)) 3> config-cc.mak -include config-cc.mak # Pauth Tests @@ -XXX,XX +XXX,XX @@ endif ifneq ($(CROSS_CC_HAS_SVE),) # System Registers Tests AARCH64_TESTS += sysregs +ifneq ($(CROSS_CC_HAS_ARMV9_SME),) +sysregs: CFLAGS+=-march=armv9-a+sme -DHAS_ARMV9_SME +else sysregs: CFLAGS+=-march=armv8.1-a+sve +endif # SVE ioctl test AARCH64_TESTS += sve-ioctls -- 2.25.1 From: Philippe Mathieu-Daudé <philmd@linaro.org> This function is not used anywhere outside this file, so we can make the function "static void". Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Eric Auger <eric.auger@redhat.com> Message-id: 20221216214924.4711-2-philmd@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/smmu-common.h | 3 --- hw/arm/smmu-common.c | 2 +- 2 files changed, 1 insertion(+), 4 deletions(-) diff --git a/include/hw/arm/smmu-common.h b/include/hw/arm/smmu-common.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/smmu-common.h +++ b/include/hw/arm/smmu-common.h @@ -XXX,XX +XXX,XX @@ void smmu_iotlb_inv_iova(SMMUState *s, int asid, dma_addr_t iova, /* Unmap the range of all the notifiers registered to any IOMMU mr */ void smmu_inv_notifiers_all(SMMUState *s); -/* Unmap the range of all the notifiers registered to @mr */ -void smmu_inv_notifiers_mr(IOMMUMemoryRegion *mr); - #endif /* HW_ARM_SMMU_COMMON_H */ diff --git a/hw/arm/smmu-common.c b/hw/arm/smmu-common.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/smmu-common.c +++ b/hw/arm/smmu-common.c @@ -XXX,XX +XXX,XX @@ static void smmu_unmap_notifier_range(IOMMUNotifier *n) } /* Unmap all notifiers attached to @mr */ -inline void smmu_inv_notifiers_mr(IOMMUMemoryRegion *mr) +static void smmu_inv_notifiers_mr(IOMMUMemoryRegion *mr) { IOMMUNotifier *n; -- 2.25.1 From: Philippe Mathieu-Daudé <philmd@linaro.org> When using Clang ("Apple clang version 14.0.0 (clang-1400.0.29.202)") and building with -Wall we get: hw/arm/smmu-common.c:173:33: warning: static function 'smmu_hash_remove_by_asid_iova' is used in an inline function with external linkage [-Wstatic-in-inline] hw/arm/smmu-common.h:170:1: note: use 'static' to give inline function 'smmu_iotlb_inv_iova' internal linkage void smmu_iotlb_inv_iova(SMMUState *s, int asid, dma_addr_t iova, ^ static None of our code base require / use inlined functions with external linkage. Some places use internal inlining in the hot path. These two functions are certainly not in any hot path and don't justify any inlining, so these are likely oversights rather than intentional. Reported-by: Stefan Weil <sw@weilnetz.de> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Eric Auger <eric.auger@redhat.com> Message-id: 20221216214924.4711-3-philmd@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/smmu-common.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) diff --git a/hw/arm/smmu-common.c b/hw/arm/smmu-common.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/smmu-common.c +++ b/hw/arm/smmu-common.c @@ -XXX,XX +XXX,XX @@ void smmu_iotlb_insert(SMMUState *bs, SMMUTransCfg *cfg, SMMUTLBEntry *new) g_hash_table_insert(bs->iotlb, key, new); } -inline void smmu_iotlb_inv_all(SMMUState *s) +void smmu_iotlb_inv_all(SMMUState *s) { trace_smmu_iotlb_inv_all(); g_hash_table_remove_all(s->iotlb); @@ -XXX,XX +XXX,XX @@ static gboolean smmu_hash_remove_by_asid_iova(gpointer key, gpointer value, ((entry->iova & ~info->mask) == info->iova); } -inline void -smmu_iotlb_inv_iova(SMMUState *s, int asid, dma_addr_t iova, - uint8_t tg, uint64_t num_pages, uint8_t ttl) +void smmu_iotlb_inv_iova(SMMUState *s, int asid, dma_addr_t iova, + uint8_t tg, uint64_t num_pages, uint8_t ttl) { /* if tg is not set we use 4KB range invalidation */ uint8_t granule = tg ? tg * 2 + 10 : 12; @@ -XXX,XX +XXX,XX @@ smmu_iotlb_inv_iova(SMMUState *s, int asid, dma_addr_t iova, &info); } -inline void smmu_iotlb_inv_asid(SMMUState *s, uint16_t asid) +void smmu_iotlb_inv_asid(SMMUState *s, uint16_t asid) { trace_smmu_iotlb_inv_asid(asid); g_hash_table_foreach_remove(s->iotlb, smmu_hash_remove_by_asid, &asid); @@ -XXX,XX +XXX,XX @@ error: * * return 0 on success */ -inline int smmu_ptw(SMMUTransCfg *cfg, dma_addr_t iova, IOMMUAccessFlags perm, - SMMUTLBEntry *tlbe, SMMUPTWEventInfo *info) +int smmu_ptw(SMMUTransCfg *cfg, dma_addr_t iova, IOMMUAccessFlags perm, + SMMUTLBEntry *tlbe, SMMUPTWEventInfo *info) { if (!cfg->aa64) { /* -- 2.25.1 From: Jean-Christophe Dubois <jcd@tribudubois.net> So far the GPT timers were unable to raise IRQs to the processor. Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/fsl-imx7.h | 5 +++++ hw/arm/fsl-imx7.c | 10 ++++++++++ 2 files changed, 15 insertions(+) diff --git a/include/hw/arm/fsl-imx7.h b/include/hw/arm/fsl-imx7.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/fsl-imx7.h +++ b/include/hw/arm/fsl-imx7.h @@ -XXX,XX +XXX,XX @@ enum FslIMX7IRQs { FSL_IMX7_USB2_IRQ = 42, FSL_IMX7_USB3_IRQ = 40, + FSL_IMX7_GPT1_IRQ = 55, + FSL_IMX7_GPT2_IRQ = 54, + FSL_IMX7_GPT3_IRQ = 53, + FSL_IMX7_GPT4_IRQ = 52, + FSL_IMX7_WDOG1_IRQ = 78, FSL_IMX7_WDOG2_IRQ = 79, FSL_IMX7_WDOG3_IRQ = 10, diff --git a/hw/arm/fsl-imx7.c b/hw/arm/fsl-imx7.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/fsl-imx7.c +++ b/hw/arm/fsl-imx7.c @@ -XXX,XX +XXX,XX @@ static void fsl_imx7_realize(DeviceState *dev, Error **errp) FSL_IMX7_GPT4_ADDR, }; + static const int FSL_IMX7_GPTn_IRQ[FSL_IMX7_NUM_GPTS] = { + FSL_IMX7_GPT1_IRQ, + FSL_IMX7_GPT2_IRQ, + FSL_IMX7_GPT3_IRQ, + FSL_IMX7_GPT4_IRQ, + }; + s->gpt[i].ccm = IMX_CCM(&s->ccm); sysbus_realize(SYS_BUS_DEVICE(&s->gpt[i]), &error_abort); sysbus_mmio_map(SYS_BUS_DEVICE(&s->gpt[i]), 0, FSL_IMX7_GPTn_ADDR[i]); + sysbus_connect_irq(SYS_BUS_DEVICE(&s->gpt[i]), 0, + qdev_get_gpio_in(DEVICE(&s->a7mpcore), + FSL_IMX7_GPTn_IRQ[i])); } for (i = 0; i < FSL_IMX7_NUM_GPIOS; i++) { -- 2.25.1 From: Jean-Christophe Dubois <jcd@tribudubois.net> CCM derived clocks will have to be added later. Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/misc/imx7_ccm.c | 49 +++++++++++++++++++++++++++++++++++++--------- 1 file changed, 40 insertions(+), 9 deletions(-) diff --git a/hw/misc/imx7_ccm.c b/hw/misc/imx7_ccm.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/imx7_ccm.c +++ b/hw/misc/imx7_ccm.c @@ -XXX,XX +XXX,XX @@ #include "hw/misc/imx7_ccm.h" #include "migration/vmstate.h" +#include "trace.h" + +#define CKIH_FREQ 24000000 /* 24MHz crystal input */ + static void imx7_analog_reset(DeviceState *dev) { IMX7AnalogState *s = IMX7_ANALOG(dev); @@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_imx7_ccm = { static uint32_t imx7_ccm_get_clock_frequency(IMXCCMState *dev, IMXClk clock) { /* - * This function is "consumed" by GPT emulation code, however on - * i.MX7 each GPT block can have their own clock root. This means - * that this functions needs somehow to know requester's identity - * and the way to pass it: be it via additional IMXClk constants - * or by adding another argument to this method needs to be - * figured out + * This function is "consumed" by GPT emulation code. Some clocks + * have fixed frequencies and we can provide requested frequency + * easily. However for CCM provided clocks (like IPG) each GPT + * timer can have its own clock root. + * This means we need additionnal information when calling this + * function to know the requester's identity. */ - qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: Not implemented\n", - TYPE_IMX7_CCM, __func__); - return 0; + uint32_t freq = 0; + + switch (clock) { + case CLK_NONE: + break; + case CLK_32k: + freq = CKIL_FREQ; + break; + case CLK_HIGH: + freq = CKIH_FREQ; + break; + case CLK_IPG: + case CLK_IPG_HIGH: + /* + * For now we don't have a way to figure out the device this + * function is called for. Until then the IPG derived clocks + * are left unimplemented. + */ + qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: Clock %d Not implemented\n", + TYPE_IMX7_CCM, __func__, clock); + break; + default: + qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: unsupported clock %d\n", + TYPE_IMX7_CCM, __func__, clock); + break; + } + + trace_ccm_clock_freq(clock, freq); + + return freq; } static void imx7_ccm_class_init(ObjectClass *klass, void *data) -- 2.25.1 From: Jean-Christophe Dubois <jcd@tribudubois.net> The i.MX6UL doesn't support CLK_HIGH ou CLK_HIGH_DIV clock source. Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/timer/imx_gpt.h | 1 + hw/arm/fsl-imx6ul.c | 2 +- hw/misc/imx6ul_ccm.c | 6 ------ hw/timer/imx_gpt.c | 25 +++++++++++++++++++++++++ 4 files changed, 27 insertions(+), 7 deletions(-) diff --git a/include/hw/timer/imx_gpt.h b/include/hw/timer/imx_gpt.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/timer/imx_gpt.h +++ b/include/hw/timer/imx_gpt.h @@ -XXX,XX +XXX,XX @@ #define TYPE_IMX25_GPT "imx25.gpt" #define TYPE_IMX31_GPT "imx31.gpt" #define TYPE_IMX6_GPT "imx6.gpt" +#define TYPE_IMX6UL_GPT "imx6ul.gpt" #define TYPE_IMX7_GPT "imx7.gpt" #define TYPE_IMX_GPT TYPE_IMX25_GPT diff --git a/hw/arm/fsl-imx6ul.c b/hw/arm/fsl-imx6ul.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/fsl-imx6ul.c +++ b/hw/arm/fsl-imx6ul.c @@ -XXX,XX +XXX,XX @@ static void fsl_imx6ul_init(Object *obj) */ for (i = 0; i < FSL_IMX6UL_NUM_GPTS; i++) { snprintf(name, NAME_SIZE, "gpt%d", i); - object_initialize_child(obj, name, &s->gpt[i], TYPE_IMX7_GPT); + object_initialize_child(obj, name, &s->gpt[i], TYPE_IMX6UL_GPT); } /* diff --git a/hw/misc/imx6ul_ccm.c b/hw/misc/imx6ul_ccm.c index XXXXXXX..XXXXXXX 100644 --- a/hw/misc/imx6ul_ccm.c +++ b/hw/misc/imx6ul_ccm.c @@ -XXX,XX +XXX,XX @@ static uint32_t imx6ul_ccm_get_clock_frequency(IMXCCMState *dev, IMXClk clock) case CLK_32k: freq = CKIL_FREQ; break; - case CLK_HIGH: - freq = CKIH_FREQ; - break; - case CLK_HIGH_DIV: - freq = CKIH_FREQ / 8; - break; default: qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: unsupported clock %d\n", TYPE_IMX6UL_CCM, __func__, clock); diff --git a/hw/timer/imx_gpt.c b/hw/timer/imx_gpt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/timer/imx_gpt.c +++ b/hw/timer/imx_gpt.c @@ -XXX,XX +XXX,XX @@ static const IMXClk imx6_gpt_clocks[] = { CLK_HIGH, /* 111 reference clock */ }; +static const IMXClk imx6ul_gpt_clocks[] = { + CLK_NONE, /* 000 No clock source */ + CLK_IPG, /* 001 ipg_clk, 532MHz*/ + CLK_IPG_HIGH, /* 010 ipg_clk_highfreq */ + CLK_EXT, /* 011 External clock */ + CLK_32k, /* 100 ipg_clk_32k */ + CLK_NONE, /* 101 not defined */ + CLK_NONE, /* 110 not defined */ + CLK_NONE, /* 111 not defined */ +}; + static const IMXClk imx7_gpt_clocks[] = { CLK_NONE, /* 000 No clock source */ CLK_IPG, /* 001 ipg_clk, 532MHz*/ @@ -XXX,XX +XXX,XX @@ static void imx6_gpt_init(Object *obj) s->clocks = imx6_gpt_clocks; } +static void imx6ul_gpt_init(Object *obj) +{ + IMXGPTState *s = IMX_GPT(obj); + + s->clocks = imx6ul_gpt_clocks; +} + static void imx7_gpt_init(Object *obj) { IMXGPTState *s = IMX_GPT(obj); @@ -XXX,XX +XXX,XX @@ static const TypeInfo imx6_gpt_info = { .instance_init = imx6_gpt_init, }; +static const TypeInfo imx6ul_gpt_info = { + .name = TYPE_IMX6UL_GPT, + .parent = TYPE_IMX25_GPT, + .instance_init = imx6ul_gpt_init, +}; + static const TypeInfo imx7_gpt_info = { .name = TYPE_IMX7_GPT, .parent = TYPE_IMX25_GPT, @@ -XXX,XX +XXX,XX @@ static void imx_gpt_register_types(void) type_register_static(&imx25_gpt_info); type_register_static(&imx31_gpt_info); type_register_static(&imx6_gpt_info); + type_register_static(&imx6ul_gpt_info); type_register_static(&imx7_gpt_info); } -- 2.25.1 From: Jean-Christophe Dubois <jcd@tribudubois.net> IRQs were not associated to the various GPIO devices inside i.MX7D. This patch brings the i.MX7D on par with i.MX6. Signed-off-by: Jean-Christophe Dubois <jcd@tribudubois.net> Message-id: 20221226101418.415170-1-jcd@tribudubois.net Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/fsl-imx7.h | 15 +++++++++++++++ hw/arm/fsl-imx7.c | 31 ++++++++++++++++++++++++++++++- 2 files changed, 45 insertions(+), 1 deletion(-) diff --git a/include/hw/arm/fsl-imx7.h b/include/hw/arm/fsl-imx7.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/fsl-imx7.h +++ b/include/hw/arm/fsl-imx7.h @@ -XXX,XX +XXX,XX @@ enum FslIMX7IRQs { FSL_IMX7_GPT3_IRQ = 53, FSL_IMX7_GPT4_IRQ = 52, + FSL_IMX7_GPIO1_LOW_IRQ = 64, + FSL_IMX7_GPIO1_HIGH_IRQ = 65, + FSL_IMX7_GPIO2_LOW_IRQ = 66, + FSL_IMX7_GPIO2_HIGH_IRQ = 67, + FSL_IMX7_GPIO3_LOW_IRQ = 68, + FSL_IMX7_GPIO3_HIGH_IRQ = 69, + FSL_IMX7_GPIO4_LOW_IRQ = 70, + FSL_IMX7_GPIO4_HIGH_IRQ = 71, + FSL_IMX7_GPIO5_LOW_IRQ = 72, + FSL_IMX7_GPIO5_HIGH_IRQ = 73, + FSL_IMX7_GPIO6_LOW_IRQ = 74, + FSL_IMX7_GPIO6_HIGH_IRQ = 75, + FSL_IMX7_GPIO7_LOW_IRQ = 76, + FSL_IMX7_GPIO7_HIGH_IRQ = 77, + FSL_IMX7_WDOG1_IRQ = 78, FSL_IMX7_WDOG2_IRQ = 79, FSL_IMX7_WDOG3_IRQ = 10, diff --git a/hw/arm/fsl-imx7.c b/hw/arm/fsl-imx7.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/fsl-imx7.c +++ b/hw/arm/fsl-imx7.c @@ -XXX,XX +XXX,XX @@ static void fsl_imx7_realize(DeviceState *dev, Error **errp) FSL_IMX7_GPIO7_ADDR, }; + static const int FSL_IMX7_GPIOn_LOW_IRQ[FSL_IMX7_NUM_GPIOS] = { + FSL_IMX7_GPIO1_LOW_IRQ, + FSL_IMX7_GPIO2_LOW_IRQ, + FSL_IMX7_GPIO3_LOW_IRQ, + FSL_IMX7_GPIO4_LOW_IRQ, + FSL_IMX7_GPIO5_LOW_IRQ, + FSL_IMX7_GPIO6_LOW_IRQ, + FSL_IMX7_GPIO7_LOW_IRQ, + }; + + static const int FSL_IMX7_GPIOn_HIGH_IRQ[FSL_IMX7_NUM_GPIOS] = { + FSL_IMX7_GPIO1_HIGH_IRQ, + FSL_IMX7_GPIO2_HIGH_IRQ, + FSL_IMX7_GPIO3_HIGH_IRQ, + FSL_IMX7_GPIO4_HIGH_IRQ, + FSL_IMX7_GPIO5_HIGH_IRQ, + FSL_IMX7_GPIO6_HIGH_IRQ, + FSL_IMX7_GPIO7_HIGH_IRQ, + }; + sysbus_realize(SYS_BUS_DEVICE(&s->gpio[i]), &error_abort); - sysbus_mmio_map(SYS_BUS_DEVICE(&s->gpio[i]), 0, FSL_IMX7_GPIOn_ADDR[i]); + sysbus_mmio_map(SYS_BUS_DEVICE(&s->gpio[i]), 0, + FSL_IMX7_GPIOn_ADDR[i]); + + sysbus_connect_irq(SYS_BUS_DEVICE(&s->gpio[i]), 0, + qdev_get_gpio_in(DEVICE(&s->a7mpcore), + FSL_IMX7_GPIOn_LOW_IRQ[i])); + + sysbus_connect_irq(SYS_BUS_DEVICE(&s->gpio[i]), 1, + qdev_get_gpio_in(DEVICE(&s->a7mpcore), + FSL_IMX7_GPIOn_HIGH_IRQ[i])); } /* -- 2.25.1 From: Stephen Longfield <slongfield@google.com> Size is used at lines 1088/1188 for the loop, which reads the last 4 bytes from the crc_ptr so it does need to get increased, however it shouldn't be increased before the buffer is passed to CRC computation, or the crc32 function will access uninitialized memory. This was pointed out to me by clg@kaod.org during the code review of a similar patch to hw/net/ftgmac100.c Change-Id: Ib0464303b191af1e28abeb2f5105eb25aadb5e9b Signed-off-by: Stephen Longfield <slongfield@google.com> Reviewed-by: Patrick Venture <venture@google.com> Message-id: 20221221183202.3788132-1-slongfield@google.com Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/net/imx_fec.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/hw/net/imx_fec.c b/hw/net/imx_fec.c index XXXXXXX..XXXXXXX 100644 --- a/hw/net/imx_fec.c +++ b/hw/net/imx_fec.c @@ -XXX,XX +XXX,XX @@ static ssize_t imx_fec_receive(NetClientState *nc, const uint8_t *buf, return 0; } - /* 4 bytes for the CRC. */ - size += 4; crc = cpu_to_be32(crc32(~0, buf, size)); + /* Increase size by 4, loop below reads the last 4 bytes from crc_ptr. */ + size += 4; crc_ptr = (uint8_t *) &crc; /* Huge frames are truncated. */ @@ -XXX,XX +XXX,XX @@ static ssize_t imx_enet_receive(NetClientState *nc, const uint8_t *buf, return 0; } - /* 4 bytes for the CRC. */ - size += 4; crc = cpu_to_be32(crc32(~0, buf, size)); + /* Increase size by 4, loop below reads the last 4 bytes from crc_ptr. */ + size += 4; crc_ptr = (uint8_t *) &crc; if (shift16) { -- 2.25.1