Most of this is the Neon decodetree patches, followed by Edgar's versal cleanups. thanks -- PMM The following changes since commit 2ef486e76d64436be90f7359a3071fb2a56ce835: Merge remote-tracking branch 'remotes/marcel/tags/rdma-pull-request' into staging (2020-05-03 14:12:56 +0100) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20200504 for you to fetch changes up to 9aefc6cf9b73f66062d2f914a0136756e7a28211: target/arm: Move gen_ function typedefs to translate.h (2020-05-04 12:59:26 +0100) ---------------------------------------------------------------- target-arm queue: * Start of conversion of Neon insns to decodetree * versal board: support SD and RTC * Implement ARMv8.2-TTS2UXN * Make VQDMULL undefined when U=1 * Some minor code cleanups ---------------------------------------------------------------- Edgar E. Iglesias (11): hw/arm: versal: Remove inclusion of arm_gicv3_common.h hw/arm: versal: Move misplaced comment hw/arm: versal-virt: Fix typo xlnx-ve -> xlnx-versal hw/arm: versal: Embed the UARTs into the SoC type hw/arm: versal: Embed the GEMs into the SoC type hw/arm: versal: Embed the ADMAs into the SoC type hw/arm: versal: Embed the APUs into the SoC type hw/arm: versal: Add support for SD hw/arm: versal: Add support for the RTC hw/arm: versal-virt: Add support for SD hw/arm: versal-virt: Add support for the RTC Fredrik Strupe (1): target/arm: Make VQDMULL undefined when U=1 Peter Maydell (25): target/arm: Don't use a TLB for ARMMMUIdx_Stage2 target/arm: Use enum constant in get_phys_addr_lpae() call target/arm: Add new 's1_is_el0' argument to get_phys_addr_lpae() target/arm: Implement ARMv8.2-TTS2UXN target/arm: Use correct variable for setting 'max' cpu's ID_AA64DFR0 target/arm/translate-vfp.inc.c: Remove duplicate simd_r32 check target/arm: Don't allow Thumb Neon insns without FEATURE_NEON target/arm: Add stubs for AArch32 Neon decodetree target/arm: Convert VCMLA (vector) to decodetree target/arm: Convert VCADD (vector) to decodetree target/arm: Convert V[US]DOT (vector) to decodetree target/arm: Convert VFM[AS]L (vector) to decodetree target/arm: Convert VCMLA (scalar) to decodetree target/arm: Convert V[US]DOT (scalar) to decodetree target/arm: Convert VFM[AS]L (scalar) to decodetree target/arm: Convert Neon load/store multiple structures to decodetree target/arm: Convert Neon 'load single structure to all lanes' to decodetree target/arm: Convert Neon 'load/store single structure' to decodetree target/arm: Convert Neon 3-reg-same VADD/VSUB to decodetree target/arm: Convert Neon 3-reg-same logic ops to decodetree target/arm: Convert Neon 3-reg-same VMAX/VMIN to decodetree target/arm: Convert Neon 3-reg-same comparisons to decodetree target/arm: Convert Neon 3-reg-same VQADD/VQSUB to decodetree target/arm: Convert Neon 3-reg-same VMUL, VMLA, VMLS, VSHL to decodetree target/arm: Move gen_ function typedefs to translate.h Philippe Mathieu-Daudé (2): hw/arm/mps2-tz: Use TYPE_IOTKIT instead of hardcoded string target/arm: Use uint64_t for midr field in CPU state struct include/hw/arm/xlnx-versal.h | 31 +- target/arm/cpu-param.h | 2 +- target/arm/cpu.h | 38 ++- target/arm/translate-a64.h | 9 - target/arm/translate.h | 26 ++ target/arm/neon-dp.decode | 86 +++++ target/arm/neon-ls.decode | 52 +++ target/arm/neon-shared.decode | 66 ++++ hw/arm/mps2-tz.c | 2 +- hw/arm/xlnx-versal-virt.c | 74 ++++- hw/arm/xlnx-versal.c | 115 +++++-- target/arm/cpu.c | 3 +- target/arm/cpu64.c | 8 +- target/arm/helper.c | 183 ++++------ target/arm/translate-a64.c | 17 - target/arm/translate-neon.inc.c | 714 +++++++++++++++++++++++++++++++++++++++ target/arm/translate-vfp.inc.c | 6 - target/arm/translate.c | 716 +++------------------------------------- target/arm/Makefile.objs | 18 + 19 files changed, 1302 insertions(+), 864 deletions(-) create mode 100644 target/arm/neon-dp.decode create mode 100644 target/arm/neon-ls.decode create mode 100644 target/arm/neon-shared.decode create mode 100644 target/arm/translate-neon.inc.c From: Fredrik Strupe <fredrik@strupe.net> According to Arm ARM, VQDMULL is only valid when U=0, while having U=1 is unallocated. Signed-off-by: Fredrik Strupe <fredrik@strupe.net> Fixes: 695272dcb976 ("target-arm: Handle UNDEF cases for Neon 3-regs-different-widths") 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) {0, 0, 0, 0}, /* VMLSL */ {0, 0, 0, 9}, /* VQDMLSL */ {0, 0, 0, 0}, /* Integer VMULL */ - {0, 0, 0, 1}, /* VQDMULL */ + {0, 0, 0, 9}, /* VQDMULL */ {0, 0, 0, 0xa}, /* Polynomial VMULL */ {0, 0, 0, 7}, /* Reserved: always UNDEF */ }; -- 2.20.1 From: Philippe Mathieu-Daudé <f4bug@amsat.org> By using the TYPE_* definitions for devices, we can: - quickly find where devices are used with 'git-grep' - easily rename a device (one-line change). Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-id: 20200428154650.21991-1-f4bug@amsat.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/mps2-tz.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/arm/mps2-tz.c b/hw/arm/mps2-tz.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/mps2-tz.c +++ b/hw/arm/mps2-tz.c @@ -XXX,XX +XXX,XX @@ static void mps2tz_common_init(MachineState *machine) exit(EXIT_FAILURE); } - sysbus_init_child_obj(OBJECT(machine), "iotkit", &mms->iotkit, + sysbus_init_child_obj(OBJECT(machine), TYPE_IOTKIT, &mms->iotkit, sizeof(mms->iotkit), mmc->armsse_type); iotkitdev = DEVICE(&mms->iotkit); object_property_set_link(OBJECT(&mms->iotkit), OBJECT(system_memory), -- 2.20.1 We define ARMMMUIdx_Stage2 as being an MMU index which uses a QEMU TLB. However we never actually use the TLB -- all stage 2 lookups are done by direct calls to get_phys_addr_lpae() followed by a physical address load via address_space_ld*(). Remove Stage2 from the list of ARM MMU indexes which correspond to real core MMU indexes, and instead put it in the set of "NOTLB" ARM MMU indexes. This allows us to drop NB_MMU_MODES to 11. It also means we can safely add support for the ARMv8.3-TTS2UXN extension, which adds permission bits to the stage 2 descriptors which define execute permission separatel for EL0 and EL1; supporting that while keeping Stage2 in a QEMU TLB would require us to use separate TLBs for "Stage2 for an EL0 access" and "Stage2 for an EL1 access", which is a lot of extra complication given we aren't even using the QEMU TLB. In the process of updating the comment on our MMU index use, fix a couple of other minor errors: * NS EL2 EL2&0 was missing from the list in the comment * some text hadn't been updated from when we bumped NB_MMU_MODES above 8 Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200330210400.11724-2-peter.maydell@linaro.org --- target/arm/cpu-param.h | 2 +- target/arm/cpu.h | 21 +++++--- target/arm/helper.c | 112 ++++------------------------------------- 3 files changed, 27 insertions(+), 108 deletions(-) diff --git a/target/arm/cpu-param.h b/target/arm/cpu-param.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cpu-param.h +++ b/target/arm/cpu-param.h @@ -XXX,XX +XXX,XX @@ # define TARGET_PAGE_BITS_MIN 10 #endif -#define NB_MMU_MODES 12 +#define NB_MMU_MODES 11 #endif 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 @@ bool write_cpustate_to_list(ARMCPU *cpu, bool kvm_sync); * handling via the TLB. The only way to do a stage 1 translation without * the immediate stage 2 translation is via the ATS or AT system insns, * which can be slow-pathed and always do a page table walk. + * The only use of stage 2 translations is either as part of an s1+2 + * lookup or when loading the descriptors during a stage 1 page table walk, + * and in both those cases we don't use the TLB. * 4. we can also safely fold together the "32 bit EL3" and "64 bit EL3" * translation regimes, because they map reasonably well to each other * and they can't both be active at the same time. @@ -XXX,XX +XXX,XX @@ bool write_cpustate_to_list(ARMCPU *cpu, bool kvm_sync); * NS EL1 EL1&0 stage 1+2 (aka NS PL1) * NS EL1 EL1&0 stage 1+2 +PAN * NS EL0 EL2&0 + * NS EL2 EL2&0 * NS EL2 EL2&0 +PAN * NS EL2 (aka NS PL2) * S EL0 EL1&0 (aka S PL0) * S EL1 EL1&0 (not used if EL3 is 32 bit) * S EL1 EL1&0 +PAN * S EL3 (aka S PL1) - * NS EL1&0 stage 2 * - * for a total of 12 different mmu_idx. + * for a total of 11 different mmu_idx. * * R profile CPUs have an MPU, but can use the same set of MMU indexes * as A profile. They only need to distinguish NS EL0 and NS EL1 (and @@ -XXX,XX +XXX,XX @@ bool write_cpustate_to_list(ARMCPU *cpu, bool kvm_sync); * are not quite the same -- different CPU types (most notably M profile * vs A/R profile) would like to use MMU indexes with different semantics, * but since we don't ever need to use all of those in a single CPU we - * can avoid setting NB_MMU_MODES to more than 8. The lower bits of + * can avoid having to set NB_MMU_MODES to "total number of A profile MMU + * modes + total number of M profile MMU modes". The lower bits of * ARMMMUIdx are the core TLB mmu index, and the higher bits are always * the same for any particular CPU. * Variables of type ARMMUIdx are always full values, and the core @@ -XXX,XX +XXX,XX @@ typedef enum ARMMMUIdx { ARMMMUIdx_SE10_1_PAN = 9 | ARM_MMU_IDX_A, ARMMMUIdx_SE3 = 10 | ARM_MMU_IDX_A, - ARMMMUIdx_Stage2 = 11 | ARM_MMU_IDX_A, - /* * These are not allocated TLBs and are used only for AT system * instructions or for the first stage of an S12 page table walk. @@ -XXX,XX +XXX,XX @@ typedef enum ARMMMUIdx { ARMMMUIdx_Stage1_E0 = 0 | ARM_MMU_IDX_NOTLB, ARMMMUIdx_Stage1_E1 = 1 | ARM_MMU_IDX_NOTLB, ARMMMUIdx_Stage1_E1_PAN = 2 | ARM_MMU_IDX_NOTLB, + /* + * Not allocated a TLB: used only for second stage of an S12 page + * table walk, or for descriptor loads during first stage of an S1 + * page table walk. Note that if we ever want to have a TLB for this + * then various TLB flush insns which currently are no-ops or flush + * only stage 1 MMU indexes will need to change to flush stage 2. + */ + ARMMMUIdx_Stage2 = 3 | ARM_MMU_IDX_NOTLB, /* * M-profile. @@ -XXX,XX +XXX,XX @@ typedef enum ARMMMUIdxBit { TO_CORE_BIT(SE10_1), TO_CORE_BIT(SE10_1_PAN), TO_CORE_BIT(SE3), - TO_CORE_BIT(Stage2), TO_CORE_BIT(MUser), TO_CORE_BIT(MPriv), 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 tlbiall_nsnh_write(CPUARMState *env, const ARMCPRegInfo *ri, tlb_flush_by_mmuidx(cs, ARMMMUIdxBit_E10_1 | ARMMMUIdxBit_E10_1_PAN | - ARMMMUIdxBit_E10_0 | - ARMMMUIdxBit_Stage2); + ARMMMUIdxBit_E10_0); } static void tlbiall_nsnh_is_write(CPUARMState *env, const ARMCPRegInfo *ri, @@ -XXX,XX +XXX,XX @@ static void tlbiall_nsnh_is_write(CPUARMState *env, const ARMCPRegInfo *ri, tlb_flush_by_mmuidx_all_cpus_synced(cs, ARMMMUIdxBit_E10_1 | ARMMMUIdxBit_E10_1_PAN | - ARMMMUIdxBit_E10_0 | - ARMMMUIdxBit_Stage2); + ARMMMUIdxBit_E10_0); } -static void tlbiipas2_write(CPUARMState *env, const ARMCPRegInfo *ri, - uint64_t value) -{ - /* Invalidate by IPA. This has to invalidate any structures that - * contain only stage 2 translation information, but does not need - * to apply to structures that contain combined stage 1 and stage 2 - * translation information. - * This must NOP if EL2 isn't implemented or SCR_EL3.NS is zero. - */ - CPUState *cs = env_cpu(env); - uint64_t pageaddr; - - if (!arm_feature(env, ARM_FEATURE_EL2) || !(env->cp15.scr_el3 & SCR_NS)) { - return; - } - - pageaddr = sextract64(value << 12, 0, 40); - - tlb_flush_page_by_mmuidx(cs, pageaddr, ARMMMUIdxBit_Stage2); -} - -static void tlbiipas2_is_write(CPUARMState *env, const ARMCPRegInfo *ri, - uint64_t value) -{ - CPUState *cs = env_cpu(env); - uint64_t pageaddr; - - if (!arm_feature(env, ARM_FEATURE_EL2) || !(env->cp15.scr_el3 & SCR_NS)) { - return; - } - - pageaddr = sextract64(value << 12, 0, 40); - - tlb_flush_page_by_mmuidx_all_cpus_synced(cs, pageaddr, - ARMMMUIdxBit_Stage2); -} static void tlbiall_hyp_write(CPUARMState *env, const ARMCPRegInfo *ri, uint64_t value) @@ -XXX,XX +XXX,XX @@ static void vttbr_write(CPUARMState *env, const ARMCPRegInfo *ri, tlb_flush_by_mmuidx(cs, ARMMMUIdxBit_E10_1 | ARMMMUIdxBit_E10_1_PAN | - ARMMMUIdxBit_E10_0 | - ARMMMUIdxBit_Stage2); + ARMMMUIdxBit_E10_0); raw_write(env, ri, value); } } @@ -XXX,XX +XXX,XX @@ static int alle1_tlbmask(CPUARMState *env) return ARMMMUIdxBit_SE10_1 | ARMMMUIdxBit_SE10_1_PAN | ARMMMUIdxBit_SE10_0; - } else if (arm_feature(env, ARM_FEATURE_EL2)) { - return ARMMMUIdxBit_E10_1 | - ARMMMUIdxBit_E10_1_PAN | - ARMMMUIdxBit_E10_0 | - ARMMMUIdxBit_Stage2; } else { return ARMMMUIdxBit_E10_1 | ARMMMUIdxBit_E10_1_PAN | @@ -XXX,XX +XXX,XX @@ static void tlbi_aa64_vae3is_write(CPUARMState *env, const ARMCPRegInfo *ri, ARMMMUIdxBit_SE3); } -static void tlbi_aa64_ipas2e1_write(CPUARMState *env, const ARMCPRegInfo *ri, - uint64_t value) -{ - /* Invalidate by IPA. This has to invalidate any structures that - * contain only stage 2 translation information, but does not need - * to apply to structures that contain combined stage 1 and stage 2 - * translation information. - * This must NOP if EL2 isn't implemented or SCR_EL3.NS is zero. - */ - ARMCPU *cpu = env_archcpu(env); - CPUState *cs = CPU(cpu); - uint64_t pageaddr; - - if (!arm_feature(env, ARM_FEATURE_EL2) || !(env->cp15.scr_el3 & SCR_NS)) { - return; - } - - pageaddr = sextract64(value << 12, 0, 48); - - tlb_flush_page_by_mmuidx(cs, pageaddr, ARMMMUIdxBit_Stage2); -} - -static void tlbi_aa64_ipas2e1is_write(CPUARMState *env, const ARMCPRegInfo *ri, - uint64_t value) -{ - CPUState *cs = env_cpu(env); - uint64_t pageaddr; - - if (!arm_feature(env, ARM_FEATURE_EL2) || !(env->cp15.scr_el3 & SCR_NS)) { - return; - } - - pageaddr = sextract64(value << 12, 0, 48); - - tlb_flush_page_by_mmuidx_all_cpus_synced(cs, pageaddr, - ARMMMUIdxBit_Stage2); -} - static CPAccessResult aa64_zva_access(CPUARMState *env, const ARMCPRegInfo *ri, bool isread) { @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v8_cp_reginfo[] = { .writefn = tlbi_aa64_vae1_write }, { .name = "TLBI_IPAS2E1IS", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 0, .opc2 = 1, - .access = PL2_W, .type = ARM_CP_NO_RAW, - .writefn = tlbi_aa64_ipas2e1is_write }, + .access = PL2_W, .type = ARM_CP_NOP }, { .name = "TLBI_IPAS2LE1IS", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 0, .opc2 = 5, - .access = PL2_W, .type = ARM_CP_NO_RAW, - .writefn = tlbi_aa64_ipas2e1is_write }, + .access = PL2_W, .type = ARM_CP_NOP }, { .name = "TLBI_ALLE1IS", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 3, .opc2 = 4, .access = PL2_W, .type = ARM_CP_NO_RAW, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v8_cp_reginfo[] = { .writefn = tlbi_aa64_alle1is_write }, { .name = "TLBI_IPAS2E1", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 4, .opc2 = 1, - .access = PL2_W, .type = ARM_CP_NO_RAW, - .writefn = tlbi_aa64_ipas2e1_write }, + .access = PL2_W, .type = ARM_CP_NOP }, { .name = "TLBI_IPAS2LE1", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 4, .opc2 = 5, - .access = PL2_W, .type = ARM_CP_NO_RAW, - .writefn = tlbi_aa64_ipas2e1_write }, + .access = PL2_W, .type = ARM_CP_NOP }, { .name = "TLBI_ALLE1", .state = ARM_CP_STATE_AA64, .opc0 = 1, .opc1 = 4, .crn = 8, .crm = 7, .opc2 = 4, .access = PL2_W, .type = ARM_CP_NO_RAW, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo v8_cp_reginfo[] = { .writefn = tlbimva_hyp_is_write }, { .name = "TLBIIPAS2", .cp = 15, .opc1 = 4, .crn = 8, .crm = 4, .opc2 = 1, - .type = ARM_CP_NO_RAW, .access = PL2_W, - .writefn = tlbiipas2_write }, + .type = ARM_CP_NOP, .access = PL2_W }, { .name = "TLBIIPAS2IS", .cp = 15, .opc1 = 4, .crn = 8, .crm = 0, .opc2 = 1, - .type = ARM_CP_NO_RAW, .access = PL2_W, - .writefn = tlbiipas2_is_write }, + .type = ARM_CP_NOP, .access = PL2_W }, { .name = "TLBIIPAS2L", .cp = 15, .opc1 = 4, .crn = 8, .crm = 4, .opc2 = 5, - .type = ARM_CP_NO_RAW, .access = PL2_W, - .writefn = tlbiipas2_write }, + .type = ARM_CP_NOP, .access = PL2_W }, { .name = "TLBIIPAS2LIS", .cp = 15, .opc1 = 4, .crn = 8, .crm = 0, .opc2 = 5, - .type = ARM_CP_NO_RAW, .access = PL2_W, - .writefn = tlbiipas2_is_write }, + .type = ARM_CP_NOP, .access = PL2_W }, /* 32 bit cache operations */ { .name = "ICIALLUIS", .cp = 15, .opc1 = 0, .crn = 7, .crm = 1, .opc2 = 0, .type = ARM_CP_NOP, .access = PL1_W, .accessfn = aa64_cacheop_pou_access }, -- 2.20.1 The access_type argument to get_phys_addr_lpae() is an MMUAccessType; use the enum constant MMU_DATA_LOAD rather than a literal 0 when we call it in S1_ptw_translate(). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200330210400.11724-3-peter.maydell@linaro.org --- target/arm/helper.c | 5 +++-- 1 file changed, 3 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 @@ static hwaddr S1_ptw_translate(CPUARMState *env, ARMMMUIdx mmu_idx, pcacheattrs = &cacheattrs; } - ret = get_phys_addr_lpae(env, addr, 0, ARMMMUIdx_Stage2, &s2pa, - &txattrs, &s2prot, &s2size, fi, pcacheattrs); + ret = get_phys_addr_lpae(env, addr, MMU_DATA_LOAD, ARMMMUIdx_Stage2, + &s2pa, &txattrs, &s2prot, &s2size, fi, + pcacheattrs); if (ret) { assert(fi->type != ARMFault_None); fi->s2addr = addr; -- 2.20.1 For ARMv8.2-TTS2UXN, the stage 2 page table walk wants to know whether the stage 1 access is for EL0 or not, because whether exec permission is given can depend on whether this is an EL0 or EL1 access. Add a new argument to get_phys_addr_lpae() so the call sites can pass this information in. Since get_phys_addr_lpae() doesn't already have a doc comment, add one so we have a place to put the documentation of the semantics of the new s1_is_el0 argument. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200330210400.11724-4-peter.maydell@linaro.org --- target/arm/helper.c | 29 ++++++++++++++++++++++++++++- 1 file changed, 28 insertions(+), 1 deletion(-) 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 bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, MMUAccessType access_type, ARMMMUIdx mmu_idx, + bool s1_is_el0, hwaddr *phys_ptr, MemTxAttrs *txattrs, int *prot, target_ulong *page_size_ptr, ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs); @@ -XXX,XX +XXX,XX @@ static hwaddr S1_ptw_translate(CPUARMState *env, ARMMMUIdx mmu_idx, } ret = get_phys_addr_lpae(env, addr, MMU_DATA_LOAD, ARMMMUIdx_Stage2, + false, &s2pa, &txattrs, &s2prot, &s2size, fi, pcacheattrs); if (ret) { @@ -XXX,XX +XXX,XX @@ static ARMVAParameters aa32_va_parameters(CPUARMState *env, uint32_t va, }; } +/** + * get_phys_addr_lpae: perform one stage of page table walk, LPAE format + * + * Returns false if the translation was successful. Otherwise, phys_ptr, attrs, + * prot and page_size may not be filled in, and the populated fsr value provides + * information on why the translation aborted, in the format of a long-format + * DFSR/IFSR fault register, with the following caveats: + * * the WnR bit is never set (the caller must do this). + * + * @env: CPUARMState + * @address: virtual address to get physical address for + * @access_type: MMU_DATA_LOAD, MMU_DATA_STORE or MMU_INST_FETCH + * @mmu_idx: MMU index indicating required translation regime + * @s1_is_el0: if @mmu_idx is ARMMMUIdx_Stage2 (so this is a stage 2 page table + * walk), must be true if this is stage 2 of a stage 1+2 walk for an + * EL0 access). If @mmu_idx is anything else, @s1_is_el0 is ignored. + * @phys_ptr: set to the physical address corresponding to the virtual address + * @attrs: set to the memory transaction attributes to use + * @prot: set to the permissions for the page containing phys_ptr + * @page_size_ptr: set to the size of the page containing phys_ptr + * @fi: set to fault info if the translation fails + * @cacheattrs: (if non-NULL) set to the cacheability/shareability attributes + */ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, MMUAccessType access_type, ARMMMUIdx mmu_idx, + bool s1_is_el0, hwaddr *phys_ptr, MemTxAttrs *txattrs, int *prot, target_ulong *page_size_ptr, ARMMMUFaultInfo *fi, ARMCacheAttrs *cacheattrs) @@ -XXX,XX +XXX,XX @@ bool get_phys_addr(CPUARMState *env, target_ulong address, /* S1 is done. Now do S2 translation. */ ret = get_phys_addr_lpae(env, ipa, access_type, ARMMMUIdx_Stage2, + mmu_idx == ARMMMUIdx_E10_0, phys_ptr, attrs, &s2_prot, page_size, fi, cacheattrs != NULL ? &cacheattrs2 : NULL); @@ -XXX,XX +XXX,XX @@ bool get_phys_addr(CPUARMState *env, target_ulong address, } if (regime_using_lpae_format(env, mmu_idx)) { - return get_phys_addr_lpae(env, address, access_type, mmu_idx, + return get_phys_addr_lpae(env, address, access_type, mmu_idx, false, phys_ptr, attrs, prot, page_size, fi, cacheattrs); } else if (regime_sctlr(env, mmu_idx) & SCTLR_XP) { -- 2.20.1 The ARMv8.2-TTS2UXN feature extends the XN field in stage 2 translation table descriptors from just bit [54] to bits [54:53], allowing stage 2 to control execution permissions separately for EL0 and EL1. Implement the new semantics of the XN field and enable the feature for our 'max' CPU. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200330210400.11724-5-peter.maydell@linaro.org --- target/arm/cpu.h | 15 +++++++++++++++ target/arm/cpu.c | 1 + target/arm/cpu64.c | 2 ++ target/arm/helper.c | 37 +++++++++++++++++++++++++++++++------ 4 files changed, 49 insertions(+), 6 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 @@ static inline bool isar_feature_aa32_ccidx(const ARMISARegisters *id) return FIELD_EX32(id->id_mmfr4, ID_MMFR4, CCIDX) != 0; } +static inline bool isar_feature_aa32_tts2uxn(const ARMISARegisters *id) +{ + return FIELD_EX32(id->id_mmfr4, ID_MMFR4, XNX) != 0; +} + /* * 64-bit feature tests via id registers. */ @@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_ccidx(const ARMISARegisters *id) return FIELD_EX64(id->id_aa64mmfr2, ID_AA64MMFR2, CCIDX) != 0; } +static inline bool isar_feature_aa64_tts2uxn(const ARMISARegisters *id) +{ + return FIELD_EX64(id->id_aa64mmfr1, ID_AA64MMFR1, XNX) != 0; +} + /* * Feature tests for "does this exist in either 32-bit or 64-bit?" */ @@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_any_ccidx(const ARMISARegisters *id) return isar_feature_aa64_ccidx(id) || isar_feature_aa32_ccidx(id); } +static inline bool isar_feature_any_tts2uxn(const ARMISARegisters *id) +{ + return isar_feature_aa64_tts2uxn(id) || isar_feature_aa32_tts2uxn(id); +} + /* * Forward to the above feature tests given an ARMCPU pointer. */ 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_max_initfn(Object *obj) t = FIELD_DP32(t, ID_MMFR4, HPDS, 1); /* AA32HPD */ t = FIELD_DP32(t, ID_MMFR4, AC2, 1); /* ACTLR2, HACTLR2 */ t = FIELD_DP32(t, ID_MMFR4, CNP, 1); /* TTCNP */ + t = FIELD_DP32(t, ID_MMFR4, XNX, 1); /* TTS2UXN */ cpu->isar.id_mmfr4 = t; } #endif 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_AA64MMFR1, VH, 1); t = FIELD_DP64(t, ID_AA64MMFR1, PAN, 2); /* ATS1E1 */ t = FIELD_DP64(t, ID_AA64MMFR1, VMIDBITS, 2); /* VMID16 */ + t = FIELD_DP64(t, ID_AA64MMFR1, XNX, 1); /* TTS2UXN */ cpu->isar.id_aa64mmfr1 = t; t = cpu->isar.id_aa64mmfr2; @@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj) u = FIELD_DP32(u, ID_MMFR4, HPDS, 1); /* AA32HPD */ u = FIELD_DP32(u, ID_MMFR4, AC2, 1); /* ACTLR2, HACTLR2 */ u = FIELD_DP32(u, ID_MMFR4, CNP, 1); /* TTCNP */ + u = FIELD_DP32(u, ID_MMFR4, XNX, 1); /* TTS2UXN */ cpu->isar.id_mmfr4 = u; u = cpu->isar.id_aa64dfr0; 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 @@ simple_ap_to_rw_prot(CPUARMState *env, ARMMMUIdx mmu_idx, int ap) * * @env: CPUARMState * @s2ap: The 2-bit stage2 access permissions (S2AP) - * @xn: XN (execute-never) bit + * @xn: XN (execute-never) bits + * @s1_is_el0: true if this is S2 of an S1+2 walk for EL0 */ -static int get_S2prot(CPUARMState *env, int s2ap, int xn) +static int get_S2prot(CPUARMState *env, int s2ap, int xn, bool s1_is_el0) { int prot = 0; @@ -XXX,XX +XXX,XX @@ static int get_S2prot(CPUARMState *env, int s2ap, int xn) if (s2ap & 2) { prot |= PAGE_WRITE; } - if (!xn) { - if (arm_el_is_aa64(env, 2) || prot & PAGE_READ) { + + if (cpu_isar_feature(any_tts2uxn, env_archcpu(env))) { + switch (xn) { + case 0: prot |= PAGE_EXEC; + break; + case 1: + if (s1_is_el0) { + prot |= PAGE_EXEC; + } + break; + case 2: + break; + case 3: + if (!s1_is_el0) { + prot |= PAGE_EXEC; + } + break; + default: + g_assert_not_reached(); + } + } else { + if (!extract32(xn, 1, 1)) { + if (arm_el_is_aa64(env, 2) || prot & PAGE_READ) { + prot |= PAGE_EXEC; + } } } return prot; @@ -XXX,XX +XXX,XX @@ static bool get_phys_addr_lpae(CPUARMState *env, target_ulong address, } ap = extract32(attrs, 4, 2); - xn = extract32(attrs, 12, 1); if (mmu_idx == ARMMMUIdx_Stage2) { ns = true; - *prot = get_S2prot(env, ap, xn); + xn = extract32(attrs, 11, 2); + *prot = get_S2prot(env, ap, xn, s1_is_el0); } else { ns = extract32(attrs, 3, 1); + xn = extract32(attrs, 12, 1); pxn = extract32(attrs, 11, 1); *prot = get_S1prot(env, mmu_idx, aarch64, ap, ns, xn, pxn); } -- 2.20.1 In aarch64_max_initfn() we update both 32-bit and 64-bit ID registers. The intended pattern is that for 64-bit ID registers we use FIELD_DP64 and the uint64_t 't' register, while 32-bit ID registers use FIELD_DP32 and the uint32_t 'u' register. For ID_AA64DFR0 we accidentally used 'u', meaning that the top 32 bits of this 64-bit ID register would end up always zero. Luckily at the moment that's what they should be anyway, so this bug has no visible effects. Use the right-sized variable. Fixes: 3bec78447a958d481991 Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Laurent Desnogues <laurent.desnogues@gmail.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-id: 20200423110915.10527-1-peter.maydell@linaro.org --- target/arm/cpu64.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) 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) u = FIELD_DP32(u, ID_MMFR4, XNX, 1); /* TTS2UXN */ cpu->isar.id_mmfr4 = u; - u = cpu->isar.id_aa64dfr0; - u = FIELD_DP64(u, ID_AA64DFR0, PMUVER, 5); /* v8.4-PMU */ - cpu->isar.id_aa64dfr0 = u; + t = cpu->isar.id_aa64dfr0; + t = FIELD_DP64(t, ID_AA64DFR0, PMUVER, 5); /* v8.4-PMU */ + cpu->isar.id_aa64dfr0 = t; u = cpu->isar.id_dfr0; u = FIELD_DP32(u, ID_DFR0, PERFMON, 5); /* v8.4-PMU */ -- 2.20.1 From: Philippe Mathieu-Daudé <f4bug@amsat.org> MIDR_EL1 is a 64-bit system register with the top 32-bit being RES0. Represent it in QEMU's ARMCPU struct with a uint64_t, not a uint32_t. This fixes an error when compiling with -Werror=conversion because we were manipulating the register value using a local uint64_t variable: target/arm/cpu64.c: In function ‘aarch64_max_initfn’: target/arm/cpu64.c:628:21: error: conversion from ‘uint64_t’ {aka ‘long unsigned int’} to ‘uint32_t’ {aka ‘unsigned int’} may change value [-Werror=conversion] 628 | cpu->midr = t; | ^ and future-proofs us against a possible future architecture change using some of the top 32 bits. Suggested-by: Laurent Desnogues <laurent.desnogues@gmail.com> Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Laurent Desnogues <laurent.desnogues@gmail.com> Message-id: 20200428172634.29707-1-f4bug@amsat.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 2 +- target/arm/cpu.c | 2 +- 2 files changed, 2 insertions(+), 2 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 @@ struct ARMCPU { uint64_t id_aa64dfr0; uint64_t id_aa64dfr1; } isar; - uint32_t midr; + uint64_t midr; uint32_t revidr; uint32_t reset_fpsid; uint32_t ctr; 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 const ARMCPUInfo arm_cpus[] = { static Property arm_cpu_properties[] = { DEFINE_PROP_BOOL("start-powered-off", ARMCPU, start_powered_off, false), DEFINE_PROP_UINT32("psci-conduit", ARMCPU, psci_conduit, 0), - DEFINE_PROP_UINT32("midr", ARMCPU, midr, 0), + DEFINE_PROP_UINT64("midr", ARMCPU, midr, 0), DEFINE_PROP_UINT64("mp-affinity", ARMCPU, mp_affinity, ARM64_AFFINITY_INVALID), DEFINE_PROP_INT32("node-id", ARMCPU, node_id, CPU_UNSET_NUMA_NODE_ID), -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Remove inclusion of arm_gicv3_common.h, this already gets included via xlnx-versal.h. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-2-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/xlnx-versal.c | 1 - 1 file changed, 1 deletion(-) diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ #include "hw/arm/boot.h" #include "kvm_arm.h" #include "hw/misc/unimp.h" -#include "hw/intc/arm_gicv3_common.h" #include "hw/arm/xlnx-versal.h" #include "hw/char/pl011.h" -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Move misplaced comment. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-3-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/xlnx-versal.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_apu_cpus(Versal *s) obj = object_new(XLNX_VERSAL_ACPU_TYPE); if (!obj) { - /* Secondary CPUs start in PSCI powered-down state */ error_report("Unable to create apu.cpu[%d] of type %s", i, XLNX_VERSAL_ACPU_TYPE); exit(EXIT_FAILURE); @@ -XXX,XX +XXX,XX @@ static void versal_create_apu_cpus(Versal *s) object_property_set_int(obj, s->cfg.psci_conduit, "psci-conduit", &error_abort); if (i) { + /* Secondary CPUs start in PSCI powered-down state */ object_property_set_bool(obj, true, "start-powered-off", &error_abort); } -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Fix typo xlnx-ve -> xlnx-versal. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-4-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/xlnx-versal-virt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/arm/xlnx-versal-virt.c b/hw/arm/xlnx-versal-virt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal-virt.c +++ b/hw/arm/xlnx-versal-virt.c @@ -XXX,XX +XXX,XX @@ static void versal_virt_init(MachineState *machine) psci_conduit = QEMU_PSCI_CONDUIT_SMC; } - sysbus_init_child_obj(OBJECT(machine), "xlnx-ve", &s->soc, + sysbus_init_child_obj(OBJECT(machine), "xlnx-versal", &s->soc, sizeof(s->soc), TYPE_XLNX_VERSAL); object_property_set_link(OBJECT(&s->soc), OBJECT(machine->ram), "ddr", &error_abort); -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Embed the UARTs into the SoC type. Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-5-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 3 ++- hw/arm/xlnx-versal.c | 12 ++++++------ 2 files changed, 8 insertions(+), 7 deletions(-) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ #include "hw/sysbus.h" #include "hw/arm/boot.h" #include "hw/intc/arm_gicv3.h" +#include "hw/char/pl011.h" #define TYPE_XLNX_VERSAL "xlnx-versal" #define XLNX_VERSAL(obj) OBJECT_CHECK(Versal, (obj), TYPE_XLNX_VERSAL) @@ -XXX,XX +XXX,XX @@ typedef struct Versal { MemoryRegion mr_ocm; struct { - SysBusDevice *uart[XLNX_VERSAL_NR_UARTS]; + PL011State uart[XLNX_VERSAL_NR_UARTS]; SysBusDevice *gem[XLNX_VERSAL_NR_GEMS]; SysBusDevice *adma[XLNX_VERSAL_NR_ADMAS]; } iou; diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ #include "kvm_arm.h" #include "hw/misc/unimp.h" #include "hw/arm/xlnx-versal.h" -#include "hw/char/pl011.h" #define XLNX_VERSAL_ACPU_TYPE ARM_CPU_TYPE_NAME("cortex-a72") #define GEM_REVISION 0x40070106 @@ -XXX,XX +XXX,XX @@ static void versal_create_uarts(Versal *s, qemu_irq *pic) DeviceState *dev; MemoryRegion *mr; - dev = qdev_create(NULL, TYPE_PL011); - s->lpd.iou.uart[i] = SYS_BUS_DEVICE(dev); + sysbus_init_child_obj(OBJECT(s), name, + &s->lpd.iou.uart[i], sizeof(s->lpd.iou.uart[i]), + TYPE_PL011); + dev = DEVICE(&s->lpd.iou.uart[i]); qdev_prop_set_chr(dev, "chardev", serial_hd(i)); - object_property_add_child(OBJECT(s), name, OBJECT(dev), &error_fatal); qdev_init_nofail(dev); - mr = sysbus_mmio_get_region(s->lpd.iou.uart[i], 0); + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(dev), 0); memory_region_add_subregion(&s->mr_ps, addrs[i], mr); - sysbus_connect_irq(s->lpd.iou.uart[i], 0, pic[irqs[i]]); + sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, pic[irqs[i]]); g_free(name); } } -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Embed the GEMs into the SoC type. Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-6-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 3 ++- hw/arm/xlnx-versal.c | 15 ++++++++------- 2 files changed, 10 insertions(+), 8 deletions(-) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ #include "hw/arm/boot.h" #include "hw/intc/arm_gicv3.h" #include "hw/char/pl011.h" +#include "hw/net/cadence_gem.h" #define TYPE_XLNX_VERSAL "xlnx-versal" #define XLNX_VERSAL(obj) OBJECT_CHECK(Versal, (obj), TYPE_XLNX_VERSAL) @@ -XXX,XX +XXX,XX @@ typedef struct Versal { struct { PL011State uart[XLNX_VERSAL_NR_UARTS]; - SysBusDevice *gem[XLNX_VERSAL_NR_GEMS]; + CadenceGEMState gem[XLNX_VERSAL_NR_GEMS]; SysBusDevice *adma[XLNX_VERSAL_NR_ADMAS]; } iou; } lpd; diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_gems(Versal *s, qemu_irq *pic) DeviceState *dev; MemoryRegion *mr; - dev = qdev_create(NULL, "cadence_gem"); - s->lpd.iou.gem[i] = SYS_BUS_DEVICE(dev); - object_property_add_child(OBJECT(s), name, OBJECT(dev), &error_fatal); + sysbus_init_child_obj(OBJECT(s), name, + &s->lpd.iou.gem[i], sizeof(s->lpd.iou.gem[i]), + TYPE_CADENCE_GEM); + dev = DEVICE(&s->lpd.iou.gem[i]); if (nd->used) { qemu_check_nic_model(nd, "cadence_gem"); qdev_set_nic_properties(dev, nd); } - object_property_set_int(OBJECT(s->lpd.iou.gem[i]), + object_property_set_int(OBJECT(dev), 2, "num-priority-queues", &error_abort); - object_property_set_link(OBJECT(s->lpd.iou.gem[i]), + object_property_set_link(OBJECT(dev), OBJECT(&s->mr_ps), "dma", &error_abort); qdev_init_nofail(dev); - mr = sysbus_mmio_get_region(s->lpd.iou.gem[i], 0); + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(dev), 0); memory_region_add_subregion(&s->mr_ps, addrs[i], mr); - sysbus_connect_irq(s->lpd.iou.gem[i], 0, pic[irqs[i]]); + sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, pic[irqs[i]]); g_free(name); } } -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Embed the ADMAs into the SoC type. Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-7-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 3 ++- hw/arm/xlnx-versal.c | 14 +++++++------- 2 files changed, 9 insertions(+), 8 deletions(-) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ #include "hw/arm/boot.h" #include "hw/intc/arm_gicv3.h" #include "hw/char/pl011.h" +#include "hw/dma/xlnx-zdma.h" #include "hw/net/cadence_gem.h" #define TYPE_XLNX_VERSAL "xlnx-versal" @@ -XXX,XX +XXX,XX @@ typedef struct Versal { struct { PL011State uart[XLNX_VERSAL_NR_UARTS]; CadenceGEMState gem[XLNX_VERSAL_NR_GEMS]; - SysBusDevice *adma[XLNX_VERSAL_NR_ADMAS]; + XlnxZDMA adma[XLNX_VERSAL_NR_ADMAS]; } iou; } lpd; diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_admas(Versal *s, qemu_irq *pic) DeviceState *dev; MemoryRegion *mr; - dev = qdev_create(NULL, "xlnx.zdma"); - s->lpd.iou.adma[i] = SYS_BUS_DEVICE(dev); - object_property_set_int(OBJECT(s->lpd.iou.adma[i]), 128, "bus-width", - &error_abort); - object_property_add_child(OBJECT(s), name, OBJECT(dev), &error_fatal); + sysbus_init_child_obj(OBJECT(s), name, + &s->lpd.iou.adma[i], sizeof(s->lpd.iou.adma[i]), + TYPE_XLNX_ZDMA); + dev = DEVICE(&s->lpd.iou.adma[i]); + object_property_set_int(OBJECT(dev), 128, "bus-width", &error_abort); qdev_init_nofail(dev); - mr = sysbus_mmio_get_region(s->lpd.iou.adma[i], 0); + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(dev), 0); memory_region_add_subregion(&s->mr_ps, MM_ADMA_CH0 + i * MM_ADMA_CH0_SIZE, mr); - sysbus_connect_irq(s->lpd.iou.adma[i], 0, pic[VERSAL_ADMA_IRQ_0 + i]); + sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, pic[VERSAL_ADMA_IRQ_0 + i]); g_free(name); } } -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Embed the APUs into the SoC type. Suggested-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-8-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 2 +- hw/arm/xlnx-versal-virt.c | 4 ++-- hw/arm/xlnx-versal.c | 19 +++++-------------- 3 files changed, 8 insertions(+), 17 deletions(-) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ typedef struct Versal { struct { struct { MemoryRegion mr; - ARMCPU *cpu[XLNX_VERSAL_NR_ACPUS]; + ARMCPU cpu[XLNX_VERSAL_NR_ACPUS]; GICv3State gic; } apu; } fpd; diff --git a/hw/arm/xlnx-versal-virt.c b/hw/arm/xlnx-versal-virt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal-virt.c +++ b/hw/arm/xlnx-versal-virt.c @@ -XXX,XX +XXX,XX @@ static void versal_virt_init(MachineState *machine) s->binfo.get_dtb = versal_virt_get_dtb; s->binfo.modify_dtb = versal_virt_modify_dtb; if (machine->kernel_filename) { - arm_load_kernel(s->soc.fpd.apu.cpu[0], machine, &s->binfo); + arm_load_kernel(&s->soc.fpd.apu.cpu[0], machine, &s->binfo); } else { - AddressSpace *as = arm_boot_address_space(s->soc.fpd.apu.cpu[0], + AddressSpace *as = arm_boot_address_space(&s->soc.fpd.apu.cpu[0], &s->binfo); /* Some boot-loaders (e.g u-boot) don't like blobs at address 0 (NULL). * Offset things by 4K. */ diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_apu_cpus(Versal *s) for (i = 0; i < ARRAY_SIZE(s->fpd.apu.cpu); i++) { Object *obj; - char *name; - - obj = object_new(XLNX_VERSAL_ACPU_TYPE); - if (!obj) { - error_report("Unable to create apu.cpu[%d] of type %s", - i, XLNX_VERSAL_ACPU_TYPE); - exit(EXIT_FAILURE); - } - - name = g_strdup_printf("apu-cpu[%d]", i); - object_property_add_child(OBJECT(s), name, obj, &error_fatal); - g_free(name); + object_initialize_child(OBJECT(s), "apu-cpu[*]", + &s->fpd.apu.cpu[i], sizeof(s->fpd.apu.cpu[i]), + XLNX_VERSAL_ACPU_TYPE, &error_abort, NULL); + obj = OBJECT(&s->fpd.apu.cpu[i]); object_property_set_int(obj, s->cfg.psci_conduit, "psci-conduit", &error_abort); if (i) { @@ -XXX,XX +XXX,XX @@ static void versal_create_apu_cpus(Versal *s) object_property_set_link(obj, OBJECT(&s->fpd.apu.mr), "memory", &error_abort); object_property_set_bool(obj, true, "realized", &error_fatal); - s->fpd.apu.cpu[i] = ARM_CPU(obj); } } @@ -XXX,XX +XXX,XX @@ static void versal_create_apu_gic(Versal *s, qemu_irq *pic) } for (i = 0; i < nr_apu_cpus; i++) { - DeviceState *cpudev = DEVICE(s->fpd.apu.cpu[i]); + DeviceState *cpudev = DEVICE(&s->fpd.apu.cpu[i]); int ppibase = XLNX_VERSAL_NR_IRQS + i * GIC_INTERNAL + GIC_NR_SGIS; qemu_irq maint_irq; int ti; -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Add support for SD. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-9-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 12 ++++++++++++ hw/arm/xlnx-versal.c | 31 +++++++++++++++++++++++++++++++ 2 files changed, 43 insertions(+) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ #include "hw/sysbus.h" #include "hw/arm/boot.h" +#include "hw/sd/sdhci.h" #include "hw/intc/arm_gicv3.h" #include "hw/char/pl011.h" #include "hw/dma/xlnx-zdma.h" @@ -XXX,XX +XXX,XX @@ #define XLNX_VERSAL_NR_UARTS 2 #define XLNX_VERSAL_NR_GEMS 2 #define XLNX_VERSAL_NR_ADMAS 8 +#define XLNX_VERSAL_NR_SDS 2 #define XLNX_VERSAL_NR_IRQS 192 typedef struct Versal { @@ -XXX,XX +XXX,XX @@ typedef struct Versal { } iou; } lpd; + /* The Platform Management Controller subsystem. */ + struct { + struct { + SDHCIState sd[XLNX_VERSAL_NR_SDS]; + } iou; + } pmc; + struct { MemoryRegion *mr_ddr; uint32_t psci_conduit; @@ -XXX,XX +XXX,XX @@ typedef struct Versal { #define VERSAL_GEM1_IRQ_0 58 #define VERSAL_GEM1_WAKE_IRQ_0 59 #define VERSAL_ADMA_IRQ_0 60 +#define VERSAL_SD0_IRQ_0 126 /* Architecturally reserved IRQs suitable for virtualization. */ #define VERSAL_RSVD_IRQ_FIRST 111 @@ -XXX,XX +XXX,XX @@ typedef struct Versal { #define MM_FPD_CRF 0xfd1a0000U #define MM_FPD_CRF_SIZE 0x140000 +#define MM_PMC_SD0 0xf1040000U +#define MM_PMC_SD0_SIZE 0x10000 #define MM_PMC_CRP 0xf1260000U #define MM_PMC_CRP_SIZE 0x10000 #endif diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_admas(Versal *s, qemu_irq *pic) } } +#define SDHCI_CAPABILITIES 0x280737ec6481 /* Same as on ZynqMP. */ +static void versal_create_sds(Versal *s, qemu_irq *pic) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(s->pmc.iou.sd); i++) { + DeviceState *dev; + MemoryRegion *mr; + + sysbus_init_child_obj(OBJECT(s), "sd[*]", + &s->pmc.iou.sd[i], sizeof(s->pmc.iou.sd[i]), + TYPE_SYSBUS_SDHCI); + dev = DEVICE(&s->pmc.iou.sd[i]); + + object_property_set_uint(OBJECT(dev), + 3, "sd-spec-version", &error_fatal); + object_property_set_uint(OBJECT(dev), SDHCI_CAPABILITIES, "capareg", + &error_fatal); + object_property_set_uint(OBJECT(dev), UHS_I, "uhs", &error_fatal); + qdev_init_nofail(dev); + + mr = sysbus_mmio_get_region(SYS_BUS_DEVICE(dev), 0); + memory_region_add_subregion(&s->mr_ps, + MM_PMC_SD0 + i * MM_PMC_SD0_SIZE, mr); + + sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, + pic[VERSAL_SD0_IRQ_0 + i * 2]); + } +} + /* This takes the board allocated linear DDR memory and creates aliases * for each split DDR range/aperture on the Versal address map. */ @@ -XXX,XX +XXX,XX @@ static void versal_realize(DeviceState *dev, Error **errp) versal_create_uarts(s, pic); versal_create_gems(s, pic); versal_create_admas(s, pic); + versal_create_sds(s, pic); versal_map_ddr(s); versal_unimp(s); -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> hw/arm: versal: Add support for the RTC. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-10-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- include/hw/arm/xlnx-versal.h | 8 ++++++++ hw/arm/xlnx-versal.c | 21 +++++++++++++++++++++ 2 files changed, 29 insertions(+) diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h index XXXXXXX..XXXXXXX 100644 --- a/include/hw/arm/xlnx-versal.h +++ b/include/hw/arm/xlnx-versal.h @@ -XXX,XX +XXX,XX @@ #include "hw/char/pl011.h" #include "hw/dma/xlnx-zdma.h" #include "hw/net/cadence_gem.h" +#include "hw/rtc/xlnx-zynqmp-rtc.h" #define TYPE_XLNX_VERSAL "xlnx-versal" #define XLNX_VERSAL(obj) OBJECT_CHECK(Versal, (obj), TYPE_XLNX_VERSAL) @@ -XXX,XX +XXX,XX @@ typedef struct Versal { struct { SDHCIState sd[XLNX_VERSAL_NR_SDS]; } iou; + + XlnxZynqMPRTC rtc; } pmc; struct { @@ -XXX,XX +XXX,XX @@ typedef struct Versal { #define VERSAL_GEM1_IRQ_0 58 #define VERSAL_GEM1_WAKE_IRQ_0 59 #define VERSAL_ADMA_IRQ_0 60 +#define VERSAL_RTC_APB_ERR_IRQ 121 #define VERSAL_SD0_IRQ_0 126 +#define VERSAL_RTC_ALARM_IRQ 142 +#define VERSAL_RTC_SECONDS_IRQ 143 /* Architecturally reserved IRQs suitable for virtualization. */ #define VERSAL_RSVD_IRQ_FIRST 111 @@ -XXX,XX +XXX,XX @@ typedef struct Versal { #define MM_PMC_SD0_SIZE 0x10000 #define MM_PMC_CRP 0xf1260000U #define MM_PMC_CRP_SIZE 0x10000 +#define MM_PMC_RTC 0xf12a0000 +#define MM_PMC_RTC_SIZE 0x10000 #endif diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal.c +++ b/hw/arm/xlnx-versal.c @@ -XXX,XX +XXX,XX @@ static void versal_create_sds(Versal *s, qemu_irq *pic) } } +static void versal_create_rtc(Versal *s, qemu_irq *pic) +{ + SysBusDevice *sbd; + MemoryRegion *mr; + + sysbus_init_child_obj(OBJECT(s), "rtc", &s->pmc.rtc, sizeof(s->pmc.rtc), + TYPE_XLNX_ZYNQMP_RTC); + sbd = SYS_BUS_DEVICE(&s->pmc.rtc); + qdev_init_nofail(DEVICE(sbd)); + + mr = sysbus_mmio_get_region(sbd, 0); + memory_region_add_subregion(&s->mr_ps, MM_PMC_RTC, mr); + + /* + * TODO: Connect the ALARM and SECONDS interrupts once our RTC model + * supports them. + */ + sysbus_connect_irq(sbd, 1, pic[VERSAL_RTC_APB_ERR_IRQ]); +} + /* This takes the board allocated linear DDR memory and creates aliases * for each split DDR range/aperture on the Versal address map. */ @@ -XXX,XX +XXX,XX @@ static void versal_realize(DeviceState *dev, Error **errp) versal_create_gems(s, pic); versal_create_admas(s, pic); versal_create_sds(s, pic); + versal_create_rtc(s, pic); versal_map_ddr(s); versal_unimp(s); -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Add support for SD. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-11-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/xlnx-versal-virt.c | 46 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+) diff --git a/hw/arm/xlnx-versal-virt.c b/hw/arm/xlnx-versal-virt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal-virt.c +++ b/hw/arm/xlnx-versal-virt.c @@ -XXX,XX +XXX,XX @@ #include "hw/arm/sysbus-fdt.h" #include "hw/arm/fdt.h" #include "cpu.h" +#include "hw/qdev-properties.h" #include "hw/arm/xlnx-versal.h" #define TYPE_XLNX_VERSAL_VIRT_MACHINE MACHINE_TYPE_NAME("xlnx-versal-virt") @@ -XXX,XX +XXX,XX @@ static void fdt_add_zdma_nodes(VersalVirt *s) } } +static void fdt_add_sd_nodes(VersalVirt *s) +{ + const char clocknames[] = "clk_xin\0clk_ahb"; + const char compat[] = "arasan,sdhci-8.9a"; + int i; + + for (i = ARRAY_SIZE(s->soc.pmc.iou.sd) - 1; i >= 0; i--) { + uint64_t addr = MM_PMC_SD0 + MM_PMC_SD0_SIZE * i; + char *name = g_strdup_printf("/sdhci@%" PRIx64, addr); + + qemu_fdt_add_subnode(s->fdt, name); + + qemu_fdt_setprop_cells(s->fdt, name, "clocks", + s->phandle.clk_25Mhz, s->phandle.clk_25Mhz); + qemu_fdt_setprop(s->fdt, name, "clock-names", + clocknames, sizeof(clocknames)); + qemu_fdt_setprop_cells(s->fdt, name, "interrupts", + GIC_FDT_IRQ_TYPE_SPI, VERSAL_SD0_IRQ_0 + i * 2, + GIC_FDT_IRQ_FLAGS_LEVEL_HI); + qemu_fdt_setprop_sized_cells(s->fdt, name, "reg", + 2, addr, 2, MM_PMC_SD0_SIZE); + qemu_fdt_setprop(s->fdt, name, "compatible", compat, sizeof(compat)); + g_free(name); + } +} + static void fdt_nop_memory_nodes(void *fdt, Error **errp) { Error *err = NULL; @@ -XXX,XX +XXX,XX @@ static void create_virtio_regions(VersalVirt *s) } } +static void sd_plugin_card(SDHCIState *sd, DriveInfo *di) +{ + BlockBackend *blk = di ? blk_by_legacy_dinfo(di) : NULL; + DeviceState *card; + + card = qdev_create(qdev_get_child_bus(DEVICE(sd), "sd-bus"), TYPE_SD_CARD); + object_property_add_child(OBJECT(sd), "card[*]", OBJECT(card), + &error_fatal); + qdev_prop_set_drive(card, "drive", blk, &error_fatal); + object_property_set_bool(OBJECT(card), true, "realized", &error_fatal); +} + static void versal_virt_init(MachineState *machine) { VersalVirt *s = XLNX_VERSAL_VIRT_MACHINE(machine); int psci_conduit = QEMU_PSCI_CONDUIT_DISABLED; + int i; /* * If the user provides an Operating System to be loaded, we expect them @@ -XXX,XX +XXX,XX @@ static void versal_virt_init(MachineState *machine) fdt_add_gic_nodes(s); fdt_add_timer_nodes(s); fdt_add_zdma_nodes(s); + fdt_add_sd_nodes(s); fdt_add_cpu_nodes(s, psci_conduit); fdt_add_clk_node(s, "/clk125", 125000000, s->phandle.clk_125Mhz); fdt_add_clk_node(s, "/clk25", 25000000, s->phandle.clk_25Mhz); @@ -XXX,XX +XXX,XX @@ static void versal_virt_init(MachineState *machine) memory_region_add_subregion_overlap(get_system_memory(), 0, &s->soc.fpd.apu.mr, 0); + /* Plugin SD cards. */ + for (i = 0; i < ARRAY_SIZE(s->soc.pmc.iou.sd); i++) { + sd_plugin_card(&s->soc.pmc.iou.sd[i], drive_get_next(IF_SD)); + } + s->binfo.ram_size = machine->ram_size; s->binfo.loader_start = 0x0; s->binfo.get_dtb = versal_virt_get_dtb; -- 2.20.1 From: "Edgar E. Iglesias" <edgar.iglesias@xilinx.com> Add support for the RTC. Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Reviewed-by: Alistair Francis <alistair.francis@wdc.com> Reviewed-by: Luc Michel <luc.michel@greensocs.com> Message-id: 20200427181649.26851-12-edgar.iglesias@gmail.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/xlnx-versal-virt.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/hw/arm/xlnx-versal-virt.c b/hw/arm/xlnx-versal-virt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/xlnx-versal-virt.c +++ b/hw/arm/xlnx-versal-virt.c @@ -XXX,XX +XXX,XX @@ static void fdt_add_sd_nodes(VersalVirt *s) } } +static void fdt_add_rtc_node(VersalVirt *s) +{ + const char compat[] = "xlnx,zynqmp-rtc"; + const char interrupt_names[] = "alarm\0sec"; + char *name = g_strdup_printf("/rtc@%x", MM_PMC_RTC); + + qemu_fdt_add_subnode(s->fdt, name); + + qemu_fdt_setprop_cells(s->fdt, name, "interrupts", + GIC_FDT_IRQ_TYPE_SPI, VERSAL_RTC_ALARM_IRQ, + GIC_FDT_IRQ_FLAGS_LEVEL_HI, + GIC_FDT_IRQ_TYPE_SPI, VERSAL_RTC_SECONDS_IRQ, + GIC_FDT_IRQ_FLAGS_LEVEL_HI); + qemu_fdt_setprop(s->fdt, name, "interrupt-names", + interrupt_names, sizeof(interrupt_names)); + qemu_fdt_setprop_sized_cells(s->fdt, name, "reg", + 2, MM_PMC_RTC, 2, MM_PMC_RTC_SIZE); + qemu_fdt_setprop(s->fdt, name, "compatible", compat, sizeof(compat)); + g_free(name); +} + static void fdt_nop_memory_nodes(void *fdt, Error **errp) { Error *err = NULL; @@ -XXX,XX +XXX,XX @@ static void versal_virt_init(MachineState *machine) fdt_add_timer_nodes(s); fdt_add_zdma_nodes(s); fdt_add_sd_nodes(s); + fdt_add_rtc_node(s); fdt_add_cpu_nodes(s, psci_conduit); fdt_add_clk_node(s, "/clk125", 125000000, s->phandle.clk_125Mhz); fdt_add_clk_node(s, "/clk25", 25000000, s->phandle.clk_25Mhz); -- 2.20.1 Somewhere along theline we accidentally added a duplicate "using D16-D31 when they don't exist" check to do_vfm_dp() (probably an artifact of a patchseries rebase). Remove it. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-id: 20200430181003.21682-2-peter.maydell@linaro.org --- target/arm/translate-vfp.inc.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/target/arm/translate-vfp.inc.c b/target/arm/translate-vfp.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-vfp.inc.c +++ b/target/arm/translate-vfp.inc.c @@ -XXX,XX +XXX,XX @@ static bool do_vfm_dp(DisasContext *s, arg_VFMA_dp *a, bool neg_n, bool neg_d) return false; } - /* UNDEF accesses to D16-D31 if they don't exist. */ - if (!dc_isar_feature(aa32_simd_r32, s) && - ((a->vd | a->vn | a->vm) & 0x10)) { - return false; - } - if (!vfp_access_check(s)) { return true; } -- 2.20.1 We were accidentally permitting decode of Thumb Neon insns even if the CPU didn't have the FEATURE_NEON bit set, because the feature check was being done before the call to disas_neon_data_insn() and disas_neon_ls_insn() in the Arm decoder but was omitted from the Thumb decoder. Push the feature bit check down into the called functions so it is done for both Arm and Thumb encodings. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Message-id: 20200430181003.21682-3-peter.maydell@linaro.org --- target/arm/translate.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) 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 int disas_neon_ls_insn(DisasContext *s, uint32_t insn) TCGv_i32 tmp2; TCGv_i64 tmp64; + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return 1; + } + /* FIXME: this access check should not take precedence over UNDEF * for invalid encodings; we will generate incorrect syndrome information * for attempts to execute invalid vfp/neon encodings with FP disabled. @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) TCGv_ptr ptr1, ptr2, ptr3; TCGv_i64 tmp64; + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return 1; + } + /* FIXME: this access check should not take precedence over UNDEF * for invalid encodings; we will generate incorrect syndrome information * for attempts to execute invalid vfp/neon encodings with FP disabled. @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) if (((insn >> 25) & 7) == 1) { /* NEON Data processing. */ - if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { - goto illegal_op; - } - if (disas_neon_data_insn(s, insn)) { goto illegal_op; } @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } if ((insn & 0x0f100000) == 0x04000000) { /* NEON load/store. */ - if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { - goto illegal_op; - } - if (disas_neon_ls_insn(s, insn)) { goto illegal_op; } -- 2.20.1 Add the infrastructure for building and invoking a decodetree decoder for the AArch32 Neon encodings. At the moment the new decoder covers nothing, so we always fall back to the existing hand-written decode. We follow the same pattern we did for the VFP decodetree conversion (commit 78e138bc1f672c145ef6ace74617d and following): code that deals with Neon will be moving gradually out to translate-neon.vfp.inc, which we #include into translate.c. In order to share the decode files between A32 and T32, we split Neon into 3 parts: * data-processing * load-store * 'shared' encodings The first two groups of instructions have similar but not identical A32 and T32 encodings, so we need to manually transform the T32 encoding into the A32 one before calling the decoder; the third group covers the Neon instructions which are identical in A32 and T32. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-4-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 29 ++++++++++++++++++++++++++ target/arm/neon-ls.decode | 29 ++++++++++++++++++++++++++ target/arm/neon-shared.decode | 27 +++++++++++++++++++++++++ target/arm/translate-neon.inc.c | 32 +++++++++++++++++++++++++++++ target/arm/translate.c | 36 +++++++++++++++++++++++++++++++-- target/arm/Makefile.objs | 18 +++++++++++++++++ 6 files changed, 169 insertions(+), 2 deletions(-) create mode 100644 target/arm/neon-dp.decode create mode 100644 target/arm/neon-ls.decode create mode 100644 target/arm/neon-shared.decode create mode 100644 target/arm/translate-neon.inc.c diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ +# AArch32 Neon data-processing instruction descriptions +# +# Copyright (c) 2020 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 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 +# + +# Encodings for Neon data processing instructions where the T32 encoding +# is a simple transformation of the A32 encoding. +# More specifically, this file covers instructions where the A32 encoding is +# 0b1111_001p_qqqq_qqqq_qqqq_qqqq_qqqq_qqqq +# and the T32 encoding is +# 0b111p_1111_qqqq_qqqq_qqqq_qqqq_qqqq_qqqq +# This file works on the A32 encoding only; calling code for T32 has to +# transform the insn into the A32 version first. diff --git a/target/arm/neon-ls.decode b/target/arm/neon-ls.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/neon-ls.decode @@ -XXX,XX +XXX,XX @@ +# AArch32 Neon load/store instruction descriptions +# +# Copyright (c) 2020 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 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 +# + +# Encodings for Neon load/store instructions where the T32 encoding +# is a simple transformation of the A32 encoding. +# More specifically, this file covers instructions where the A32 encoding is +# 0b1111_0100_xxx0_xxxx_xxxx_xxxx_xxxx_xxxx +# and the T32 encoding is +# 0b1111_1001_xxx0_xxxx_xxxx_xxxx_xxxx_xxxx +# This file works on the A32 encoding only; calling code for T32 has to +# transform the insn into the A32 version first. diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ +# AArch32 Neon instruction descriptions +# +# Copyright (c) 2020 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 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 +# + +# Encodings for Neon instructions whose encoding is the same for +# both A32 and T32. + +# More specifically, this covers: +# 2reg scalar ext: 0b1111_1110_xxxx_xxxx_xxxx_1x0x_xxxx_xxxx +# 3same ext: 0b1111_110x_xxxx_xxxx_xxxx_1x0x_xxxx_xxxx diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ +/* + * ARM translation: AArch32 Neon instructions + * + * Copyright (c) 2003 Fabrice Bellard + * Copyright (c) 2005-2007 CodeSourcery + * Copyright (c) 2007 OpenedHand, Ltd. + * Copyright (c) 2020 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 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 intended to be included from translate.c; it uses + * some macros and definitions provided by that file. + * It might be possible to convert it to a standalone .c file eventually. + */ + +/* Include the generated Neon decoder */ +#include "decode-neon-dp.inc.c" +#include "decode-neon-ls.inc.c" +#include "decode-neon-shared.inc.c" 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 TCGv_ptr vfp_reg_ptr(bool dp, int reg) #define ARM_CP_RW_BIT (1 << 20) -/* Include the VFP decoder */ +/* Include the VFP and Neon decoders */ #include "translate-vfp.inc.c" +#include "translate-neon.inc.c" static inline void iwmmxt_load_reg(TCGv_i64 var, int reg) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) /* Unconditional instructions. */ /* TODO: Perhaps merge these into one decodetree output file. */ if (disas_a32_uncond(s, insn) || - disas_vfp_uncond(s, insn)) { + disas_vfp_uncond(s, insn) || + disas_neon_dp(s, insn) || + disas_neon_ls(s, insn) || + disas_neon_shared(s, insn)) { return; } /* fall back to legacy decoder */ @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) ARCH(6T2); } + if ((insn & 0xef000000) == 0xef000000) { + /* + * T32 encodings 0b111p_1111_qqqq_qqqq_qqqq_qqqq_qqqq_qqqq + * transform into + * A32 encodings 0b1111_001p_qqqq_qqqq_qqqq_qqqq_qqqq_qqqq + */ + uint32_t a32_insn = (insn & 0xe2ffffff) | + ((insn & (1 << 28)) >> 4) | (1 << 28); + + if (disas_neon_dp(s, a32_insn)) { + return; + } + } + + if ((insn & 0xff100000) == 0xf9000000) { + /* + * T32 encodings 0b1111_1001_ppp0_qqqq_qqqq_qqqq_qqqq_qqqq + * transform into + * A32 encodings 0b1111_0100_ppp0_qqqq_qqqq_qqqq_qqqq_qqqq + */ + uint32_t a32_insn = (insn & 0x00ffffff) | 0xf4000000; + + if (disas_neon_ls(s, a32_insn)) { + return; + } + } + /* * TODO: Perhaps merge these into one decodetree output file. * Note disas_vfp is written for a32 with cond field in the @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) */ if (disas_t32(s, insn) || disas_vfp_uncond(s, insn) || + disas_neon_shared(s, insn) || ((insn >> 28) == 0xe && disas_vfp(s, insn))) { return; } diff --git a/target/arm/Makefile.objs b/target/arm/Makefile.objs index XXXXXXX..XXXXXXX 100644 --- a/target/arm/Makefile.objs +++ b/target/arm/Makefile.objs @@ -XXX,XX +XXX,XX @@ target/arm/decode-sve.inc.c: $(SRC_PATH)/target/arm/sve.decode $(DECODETREE) $(PYTHON) $(DECODETREE) --decode disas_sve -o $@ $<,\ "GEN", $(TARGET_DIR)$@) +target/arm/decode-neon-shared.inc.c: $(SRC_PATH)/target/arm/neon-shared.decode $(DECODETREE) + $(call quiet-command,\ + $(PYTHON) $(DECODETREE) --static-decode disas_neon_shared -o $@ $<,\ + "GEN", $(TARGET_DIR)$@) + +target/arm/decode-neon-dp.inc.c: $(SRC_PATH)/target/arm/neon-dp.decode $(DECODETREE) + $(call quiet-command,\ + $(PYTHON) $(DECODETREE) --static-decode disas_neon_dp -o $@ $<,\ + "GEN", $(TARGET_DIR)$@) + +target/arm/decode-neon-ls.inc.c: $(SRC_PATH)/target/arm/neon-ls.decode $(DECODETREE) + $(call quiet-command,\ + $(PYTHON) $(DECODETREE) --static-decode disas_neon_ls -o $@ $<,\ + "GEN", $(TARGET_DIR)$@) + target/arm/decode-vfp.inc.c: $(SRC_PATH)/target/arm/vfp.decode $(DECODETREE) $(call quiet-command,\ $(PYTHON) $(DECODETREE) --static-decode disas_vfp -o $@ $<,\ @@ -XXX,XX +XXX,XX @@ target/arm/decode-t16.inc.c: $(SRC_PATH)/target/arm/t16.decode $(DECODETREE) "GEN", $(TARGET_DIR)$@) target/arm/translate-sve.o: target/arm/decode-sve.inc.c +target/arm/translate.o: target/arm/decode-neon-shared.inc.c +target/arm/translate.o: target/arm/decode-neon-dp.inc.c +target/arm/translate.o: target/arm/decode-neon-ls.inc.c target/arm/translate.o: target/arm/decode-vfp.inc.c target/arm/translate.o: target/arm/decode-vfp-uncond.inc.c target/arm/translate.o: target/arm/decode-a32.inc.c -- 2.20.1 Convert the VCMLA (vector) insns in the 3same extension group to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-5-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 11 ++++++++++ target/arm/translate-neon.inc.c | 37 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 11 +--------- 3 files changed, 49 insertions(+), 10 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ # More specifically, this covers: # 2reg scalar ext: 0b1111_1110_xxxx_xxxx_xxxx_1x0x_xxxx_xxxx # 3same ext: 0b1111_110x_xxxx_xxxx_xxxx_1x0x_xxxx_xxxx + +# VFP/Neon register fields; same as vfp.decode +%vm_dp 5:1 0:4 +%vm_sp 0:4 5:1 +%vn_dp 7:1 16:4 +%vn_sp 16:4 7:1 +%vd_dp 22:1 12:4 +%vd_sp 12:4 22:1 + +VCMLA 1111 110 rot:2 . 1 size:1 .... .... 1000 . q:1 . 0 .... \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ #include "decode-neon-dp.inc.c" #include "decode-neon-ls.inc.c" #include "decode-neon-shared.inc.c" + +static bool trans_VCMLA(DisasContext *s, arg_VCMLA *a) +{ + int opr_sz; + TCGv_ptr fpst; + gen_helper_gvec_3_ptr *fn_gvec_ptr; + + if (!dc_isar_feature(aa32_vcma, s) + || (!a->size && !dc_isar_feature(aa32_fp16_arith, s))) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } + + if ((a->vn | a->vm | a->vd) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + opr_sz = (1 + a->q) * 8; + fpst = get_fpstatus_ptr(1); + fn_gvec_ptr = a->size ? gen_helper_gvec_fcmlas : gen_helper_gvec_fcmlah; + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->vm), + fpst, opr_sz, opr_sz, a->rot, + fn_gvec_ptr); + tcg_temp_free_ptr(fpst); + return true; +} 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 int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) bool is_long = false, q = extract32(insn, 6, 1); bool ptr_is_env = false; - if ((insn & 0xfe200f10) == 0xfc200800) { - /* VCMLA -- 1111 110R R.1S .... .... 1000 ...0 .... */ - int size = extract32(insn, 20, 1); - data = extract32(insn, 23, 2); /* rot */ - if (!dc_isar_feature(aa32_vcma, s) - || (!size && !dc_isar_feature(aa32_fp16_arith, s))) { - return 1; - } - fn_gvec_ptr = size ? gen_helper_gvec_fcmlas : gen_helper_gvec_fcmlah; - } else if ((insn & 0xfea00f10) == 0xfc800800) { + if ((insn & 0xfea00f10) == 0xfc800800) { /* VCADD -- 1111 110R 1.0S .... .... 1000 ...0 .... */ int size = extract32(insn, 20, 1); data = extract32(insn, 24, 1); /* rot */ -- 2.20.1 Convert the VCADD (vector) insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-6-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 3 +++ target/arm/translate-neon.inc.c | 37 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 11 +--------- 3 files changed, 41 insertions(+), 10 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VCMLA 1111 110 rot:2 . 1 size:1 .... .... 1000 . q:1 . 0 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp + +VCADD 1111 110 rot:1 1 . 0 size:1 .... .... 1000 . q:1 . 0 .... \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VCMLA(DisasContext *s, arg_VCMLA *a) tcg_temp_free_ptr(fpst); return true; } + +static bool trans_VCADD(DisasContext *s, arg_VCADD *a) +{ + int opr_sz; + TCGv_ptr fpst; + gen_helper_gvec_3_ptr *fn_gvec_ptr; + + if (!dc_isar_feature(aa32_vcma, s) + || (!a->size && !dc_isar_feature(aa32_fp16_arith, s))) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } + + if ((a->vn | a->vm | a->vd) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + opr_sz = (1 + a->q) * 8; + fpst = get_fpstatus_ptr(1); + fn_gvec_ptr = a->size ? gen_helper_gvec_fcadds : gen_helper_gvec_fcaddh; + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->vm), + fpst, opr_sz, opr_sz, a->rot, + fn_gvec_ptr); + tcg_temp_free_ptr(fpst); + return true; +} 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 int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) bool is_long = false, q = extract32(insn, 6, 1); bool ptr_is_env = false; - if ((insn & 0xfea00f10) == 0xfc800800) { - /* VCADD -- 1111 110R 1.0S .... .... 1000 ...0 .... */ - int size = extract32(insn, 20, 1); - data = extract32(insn, 24, 1); /* rot */ - if (!dc_isar_feature(aa32_vcma, s) - || (!size && !dc_isar_feature(aa32_fp16_arith, s))) { - return 1; - } - fn_gvec_ptr = size ? gen_helper_gvec_fcadds : gen_helper_gvec_fcaddh; - } else if ((insn & 0xfeb00f00) == 0xfc200d00) { + if ((insn & 0xfeb00f00) == 0xfc200d00) { /* V[US]DOT -- 1111 1100 0.10 .... .... 1101 .Q.U .... */ bool u = extract32(insn, 4, 1); if (!dc_isar_feature(aa32_dp, s)) { -- 2.20.1 Convert the V[US]DOT (vector) insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-7-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 4 ++++ target/arm/translate-neon.inc.c | 32 ++++++++++++++++++++++++++++++++ target/arm/translate.c | 9 +-------- 3 files changed, 37 insertions(+), 8 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VCMLA 1111 110 rot:2 . 1 size:1 .... .... 1000 . q:1 . 0 .... \ VCADD 1111 110 rot:1 1 . 0 size:1 .... .... 1000 . q:1 . 0 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp + +# VUDOT and VSDOT +VDOT 1111 110 00 . 10 .... .... 1101 . q:1 . u:1 .... \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VCADD(DisasContext *s, arg_VCADD *a) tcg_temp_free_ptr(fpst); return true; } + +static bool trans_VDOT(DisasContext *s, arg_VDOT *a) +{ + int opr_sz; + gen_helper_gvec_3 *fn_gvec; + + if (!dc_isar_feature(aa32_dp, s)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } + + if ((a->vn | a->vm | a->vd) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + opr_sz = (1 + a->q) * 8; + fn_gvec = a->u ? gen_helper_gvec_udot_b : gen_helper_gvec_sdot_b; + tcg_gen_gvec_3_ool(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->vm), + opr_sz, opr_sz, 0, fn_gvec); + return true; +} 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 int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) bool is_long = false, q = extract32(insn, 6, 1); bool ptr_is_env = false; - if ((insn & 0xfeb00f00) == 0xfc200d00) { - /* V[US]DOT -- 1111 1100 0.10 .... .... 1101 .Q.U .... */ - bool u = extract32(insn, 4, 1); - if (!dc_isar_feature(aa32_dp, s)) { - return 1; - } - fn_gvec = u ? gen_helper_gvec_udot_b : gen_helper_gvec_sdot_b; - } else if ((insn & 0xff300f10) == 0xfc200810) { + if ((insn & 0xff300f10) == 0xfc200810) { /* VFM[AS]L -- 1111 1100 S.10 .... .... 1000 .Q.1 .... */ int is_s = extract32(insn, 23, 1); if (!dc_isar_feature(aa32_fhm, s)) { -- 2.20.1 Convert the VFM[AS]L (vector) insns to decodetree. This is the last insn in the legacy decoder for the 3same_ext group, so we can delete the legacy decoder function for the group entirely. Note that in disas_thumb2_insn() the parts of this encoding space where the decodetree decoder returns false will correctly be directed to illegal_op by the "(insn & (1 << 28))" check so they won't fall into disas_coproc_insn() by mistake. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-8-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 6 +++ target/arm/translate-neon.inc.c | 31 +++++++++++ target/arm/translate.c | 92 +-------------------------------- 3 files changed, 38 insertions(+), 91 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VCADD 1111 110 rot:1 1 . 0 size:1 .... .... 1000 . q:1 . 0 .... \ # VUDOT and VSDOT VDOT 1111 110 00 . 10 .... .... 1101 . q:1 . u:1 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp + +# VFM[AS]L +VFML 1111 110 0 s:1 . 10 .... .... 1000 . 0 . 1 .... \ + vm=%vm_sp vn=%vn_sp vd=%vd_dp q=0 +VFML 1111 110 0 s:1 . 10 .... .... 1000 . 1 . 1 .... \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp q=1 diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VDOT(DisasContext *s, arg_VDOT *a) opr_sz, opr_sz, 0, fn_gvec); return true; } + +static bool trans_VFML(DisasContext *s, arg_VFML *a) +{ + int opr_sz; + + if (!dc_isar_feature(aa32_fhm, s)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + (a->vd & 0x10)) { + return false; + } + + if (a->vd & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + opr_sz = (1 + a->q) * 8; + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(a->q, a->vn), + vfp_reg_offset(a->q, a->vm), + cpu_env, opr_sz, opr_sz, a->s, /* is_2 == 0 */ + gen_helper_gvec_fmlal_a32); + return true; +} 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) return 0; } -/* Advanced SIMD three registers of the same length extension. - * 31 25 23 22 20 16 12 11 10 9 8 3 0 - * +---------------+-----+---+-----+----+----+---+----+---+----+---------+----+ - * | 1 1 1 1 1 1 0 | op1 | D | op2 | Vn | Vd | 1 | o3 | 0 | o4 | N Q M U | Vm | - * +---------------+-----+---+-----+----+----+---+----+---+----+---------+----+ - */ -static int disas_neon_insn_3same_ext(DisasContext *s, uint32_t insn) -{ - gen_helper_gvec_3 *fn_gvec = NULL; - gen_helper_gvec_3_ptr *fn_gvec_ptr = NULL; - int rd, rn, rm, opr_sz; - int data = 0; - int off_rn, off_rm; - bool is_long = false, q = extract32(insn, 6, 1); - bool ptr_is_env = false; - - if ((insn & 0xff300f10) == 0xfc200810) { - /* VFM[AS]L -- 1111 1100 S.10 .... .... 1000 .Q.1 .... */ - int is_s = extract32(insn, 23, 1); - if (!dc_isar_feature(aa32_fhm, s)) { - return 1; - } - is_long = true; - data = is_s; /* is_2 == 0 */ - fn_gvec_ptr = gen_helper_gvec_fmlal_a32; - ptr_is_env = true; - } else { - return 1; - } - - VFP_DREG_D(rd, insn); - if (rd & q) { - return 1; - } - if (q || !is_long) { - VFP_DREG_N(rn, insn); - VFP_DREG_M(rm, insn); - if ((rn | rm) & q & !is_long) { - return 1; - } - off_rn = vfp_reg_offset(1, rn); - off_rm = vfp_reg_offset(1, rm); - } else { - rn = VFP_SREG_N(insn); - rm = VFP_SREG_M(insn); - off_rn = vfp_reg_offset(0, rn); - off_rm = vfp_reg_offset(0, rm); - } - - if (s->fp_excp_el) { - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, - syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); - return 0; - } - if (!s->vfp_enabled) { - return 1; - } - - opr_sz = (1 + q) * 8; - if (fn_gvec_ptr) { - TCGv_ptr ptr; - if (ptr_is_env) { - ptr = cpu_env; - } else { - ptr = get_fpstatus_ptr(1); - } - tcg_gen_gvec_3_ptr(vfp_reg_offset(1, rd), off_rn, off_rm, ptr, - opr_sz, opr_sz, data, fn_gvec_ptr); - if (!ptr_is_env) { - tcg_temp_free_ptr(ptr); - } - } else { - tcg_gen_gvec_3_ool(vfp_reg_offset(1, rd), off_rn, off_rm, - opr_sz, opr_sz, data, fn_gvec); - } - return 0; -} - /* Advanced SIMD two registers and a scalar extension. * 31 24 23 22 20 16 12 11 10 9 8 3 0 * +-----------------+----+---+----+----+----+---+----+---+----+---------+----+ @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } } } - } else if ((insn & 0x0e000a00) == 0x0c000800 - && arm_dc_feature(s, ARM_FEATURE_V8)) { - if (disas_neon_insn_3same_ext(s, insn)) { - goto illegal_op; - } - return; } else if ((insn & 0x0f000a00) == 0x0e000800 && arm_dc_feature(s, ARM_FEATURE_V8)) { if (disas_neon_insn_2reg_scalar_ext(s, insn)) { @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } break; } - if ((insn & 0xfe000a00) == 0xfc000800 + if ((insn & 0xff000a00) == 0xfe000800 && arm_dc_feature(s, ARM_FEATURE_V8)) { /* The Thumb2 and ARM encodings are identical. */ - if (disas_neon_insn_3same_ext(s, insn)) { - goto illegal_op; - } - } else if ((insn & 0xff000a00) == 0xfe000800 - && arm_dc_feature(s, ARM_FEATURE_V8)) { - /* The Thumb2 and ARM encodings are identical. */ if (disas_neon_insn_2reg_scalar_ext(s, insn)) { goto illegal_op; } -- 2.20.1 Convert VCMLA (scalar) in the 2reg-scalar-ext group to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-9-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 5 +++++ target/arm/translate-neon.inc.c | 40 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 26 +-------------------- 3 files changed, 46 insertions(+), 25 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VFML 1111 110 0 s:1 . 10 .... .... 1000 . 0 . 1 .... \ vm=%vm_sp vn=%vn_sp vd=%vd_dp q=0 VFML 1111 110 0 s:1 . 10 .... .... 1000 . 1 . 1 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp q=1 + +VCMLA_scalar 1111 1110 0 . rot:2 .... .... 1000 . q:1 index:1 0 vm:4 \ + vn=%vn_dp vd=%vd_dp size=0 +VCMLA_scalar 1111 1110 1 . rot:2 .... .... 1000 . q:1 . 0 .... \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp size=1 index=0 diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VFML(DisasContext *s, arg_VFML *a) gen_helper_gvec_fmlal_a32); return true; } + +static bool trans_VCMLA_scalar(DisasContext *s, arg_VCMLA_scalar *a) +{ + gen_helper_gvec_3_ptr *fn_gvec_ptr; + int opr_sz; + TCGv_ptr fpst; + + if (!dc_isar_feature(aa32_vcma, s)) { + return false; + } + if (a->size == 0 && !dc_isar_feature(aa32_fp16_arith, s)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } + + if ((a->vd | a->vn) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + fn_gvec_ptr = (a->size ? gen_helper_gvec_fcmlas_idx + : gen_helper_gvec_fcmlah_idx); + opr_sz = (1 + a->q) * 8; + fpst = get_fpstatus_ptr(1); + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->vm), + fpst, opr_sz, opr_sz, + (a->index << 2) | a->rot, fn_gvec_ptr); + tcg_temp_free_ptr(fpst); + return true; +} 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 int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) bool is_long = false, q = extract32(insn, 6, 1); bool ptr_is_env = false; - if ((insn & 0xff000f10) == 0xfe000800) { - /* VCMLA (indexed) -- 1111 1110 S.RR .... .... 1000 ...0 .... */ - int rot = extract32(insn, 20, 2); - int size = extract32(insn, 23, 1); - int index; - - if (!dc_isar_feature(aa32_vcma, s)) { - return 1; - } - if (size == 0) { - if (!dc_isar_feature(aa32_fp16_arith, s)) { - return 1; - } - /* For fp16, rm is just Vm, and index is M. */ - rm = extract32(insn, 0, 4); - index = extract32(insn, 5, 1); - } else { - /* For fp32, rm is the usual M:Vm, and index is 0. */ - VFP_DREG_M(rm, insn); - index = 0; - } - data = (index << 2) | rot; - fn_gvec_ptr = (size ? gen_helper_gvec_fcmlas_idx - : gen_helper_gvec_fcmlah_idx); - } else if ((insn & 0xffb00f00) == 0xfe200d00) { + if ((insn & 0xffb00f00) == 0xfe200d00) { /* V[US]DOT -- 1111 1110 0.10 .... .... 1101 .Q.U .... */ int u = extract32(insn, 4, 1); -- 2.20.1 Convert the V[US]DOT (scalar) insns in the 2reg-scalar-ext group to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-10-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 3 +++ target/arm/translate-neon.inc.c | 35 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 13 +----------- 3 files changed, 39 insertions(+), 12 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VCMLA_scalar 1111 1110 0 . rot:2 .... .... 1000 . q:1 index:1 0 vm:4 \ vn=%vn_dp vd=%vd_dp size=0 VCMLA_scalar 1111 1110 1 . rot:2 .... .... 1000 . q:1 . 0 .... \ vm=%vm_dp vn=%vn_dp vd=%vd_dp size=1 index=0 + +VDOT_scalar 1111 1110 0 . 10 .... .... 1101 . q:1 index:1 u:1 rm:4 \ + vm=%vm_dp vn=%vn_dp vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VCMLA_scalar(DisasContext *s, arg_VCMLA_scalar *a) tcg_temp_free_ptr(fpst); return true; } + +static bool trans_VDOT_scalar(DisasContext *s, arg_VDOT_scalar *a) +{ + gen_helper_gvec_3 *fn_gvec; + int opr_sz; + TCGv_ptr fpst; + + if (!dc_isar_feature(aa32_dp, s)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn) & 0x10)) { + return false; + } + + if ((a->vd | a->vn) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + fn_gvec = a->u ? gen_helper_gvec_udot_idx_b : gen_helper_gvec_sdot_idx_b; + opr_sz = (1 + a->q) * 8; + fpst = get_fpstatus_ptr(1); + tcg_gen_gvec_3_ool(vfp_reg_offset(1, a->vd), + vfp_reg_offset(1, a->vn), + vfp_reg_offset(1, a->rm), + opr_sz, opr_sz, a->index, fn_gvec); + tcg_temp_free_ptr(fpst); + return true; +} 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 int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) bool is_long = false, q = extract32(insn, 6, 1); bool ptr_is_env = false; - if ((insn & 0xffb00f00) == 0xfe200d00) { - /* V[US]DOT -- 1111 1110 0.10 .... .... 1101 .Q.U .... */ - int u = extract32(insn, 4, 1); - - if (!dc_isar_feature(aa32_dp, s)) { - return 1; - } - fn_gvec = u ? gen_helper_gvec_udot_idx_b : gen_helper_gvec_sdot_idx_b; - /* rm is just Vm, and index is M. */ - data = extract32(insn, 5, 1); /* index */ - rm = extract32(insn, 0, 4); - } else if ((insn & 0xffa00f10) == 0xfe000810) { + if ((insn & 0xffa00f10) == 0xfe000810) { /* VFM[AS]L -- 1111 1110 0.0S .... .... 1000 .Q.1 .... */ int is_s = extract32(insn, 20, 1); int vm20 = extract32(insn, 0, 3); -- 2.20.1 Convert the VFM[AS]L (scalar) insns in the 2reg-scalar-ext group to decodetree. These are the last ones in the group so we can remove all the legacy decode for the group. Note that in disas_thumb2_insn() the parts of this encoding space where the decodetree decoder returns false will correctly be directed to illegal_op by the "(insn & (1 << 28))" check so they won't fall into disas_coproc_insn() by mistake. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-11-peter.maydell@linaro.org --- target/arm/neon-shared.decode | 7 +++ target/arm/translate-neon.inc.c | 32 ++++++++++ target/arm/translate.c | 107 +------------------------------- 3 files changed, 40 insertions(+), 106 deletions(-) diff --git a/target/arm/neon-shared.decode b/target/arm/neon-shared.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-shared.decode +++ b/target/arm/neon-shared.decode @@ -XXX,XX +XXX,XX @@ VCMLA_scalar 1111 1110 1 . rot:2 .... .... 1000 . q:1 . 0 .... \ VDOT_scalar 1111 1110 0 . 10 .... .... 1101 . q:1 index:1 u:1 rm:4 \ vm=%vm_dp vn=%vn_dp vd=%vd_dp + +%vfml_scalar_q0_rm 0:3 5:1 +%vfml_scalar_q1_index 5:1 3:1 +VFML_scalar 1111 1110 0 . 0 s:1 .... .... 1000 . 0 . 1 index:1 ... \ + rm=%vfml_scalar_q0_rm vn=%vn_sp vd=%vd_dp q=0 +VFML_scalar 1111 1110 0 . 0 s:1 .... .... 1000 . 1 . 1 . rm:3 \ + index=%vfml_scalar_q1_index vn=%vn_dp vd=%vd_dp q=1 diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VDOT_scalar(DisasContext *s, arg_VDOT_scalar *a) tcg_temp_free_ptr(fpst); return true; } + +static bool trans_VFML_scalar(DisasContext *s, arg_VFML_scalar *a) +{ + int opr_sz; + + if (!dc_isar_feature(aa32_fhm, s)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd & 0x10) || (a->q && (a->vn & 0x10)))) { + return false; + } + + if (a->vd & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + opr_sz = (1 + a->q) * 8; + tcg_gen_gvec_3_ptr(vfp_reg_offset(1, a->vd), + vfp_reg_offset(a->q, a->vn), + vfp_reg_offset(a->q, a->rm), + cpu_env, opr_sz, opr_sz, + (a->index << 2) | a->s, /* is_2 == 0 */ + gen_helper_gvec_fmlal_idx_a32); + return true; +} 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 int disas_dsp_insn(DisasContext *s, uint32_t insn) } #define VFP_REG_SHR(x, n) (((n) > 0) ? (x) >> (n) : (x) << -(n)) -#define VFP_SREG(insn, bigbit, smallbit) \ - ((VFP_REG_SHR(insn, bigbit - 1) & 0x1e) | (((insn) >> (smallbit)) & 1)) #define VFP_DREG(reg, insn, bigbit, smallbit) do { \ if (dc_isar_feature(aa32_simd_r32, s)) { \ reg = (((insn) >> (bigbit)) & 0x0f) \ @@ -XXX,XX +XXX,XX @@ static int disas_dsp_insn(DisasContext *s, uint32_t insn) reg = ((insn) >> (bigbit)) & 0x0f; \ }} while (0) -#define VFP_SREG_D(insn) VFP_SREG(insn, 12, 22) #define VFP_DREG_D(reg, insn) VFP_DREG(reg, insn, 12, 22) -#define VFP_SREG_N(insn) VFP_SREG(insn, 16, 7) #define VFP_DREG_N(reg, insn) VFP_DREG(reg, insn, 16, 7) -#define VFP_SREG_M(insn) VFP_SREG(insn, 0, 5) #define VFP_DREG_M(reg, insn) VFP_DREG(reg, insn, 0, 5) static void gen_neon_dup_low16(TCGv_i32 var) @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) return 0; } -/* Advanced SIMD two registers and a scalar extension. - * 31 24 23 22 20 16 12 11 10 9 8 3 0 - * +-----------------+----+---+----+----+----+---+----+---+----+---------+----+ - * | 1 1 1 1 1 1 1 0 | o1 | D | o2 | Vn | Vd | 1 | o3 | 0 | o4 | N Q M U | Vm | - * +-----------------+----+---+----+----+----+---+----+---+----+---------+----+ - * - */ - -static int disas_neon_insn_2reg_scalar_ext(DisasContext *s, uint32_t insn) -{ - gen_helper_gvec_3 *fn_gvec = NULL; - gen_helper_gvec_3_ptr *fn_gvec_ptr = NULL; - int rd, rn, rm, opr_sz, data; - int off_rn, off_rm; - bool is_long = false, q = extract32(insn, 6, 1); - bool ptr_is_env = false; - - if ((insn & 0xffa00f10) == 0xfe000810) { - /* VFM[AS]L -- 1111 1110 0.0S .... .... 1000 .Q.1 .... */ - int is_s = extract32(insn, 20, 1); - int vm20 = extract32(insn, 0, 3); - int vm3 = extract32(insn, 3, 1); - int m = extract32(insn, 5, 1); - int index; - - if (!dc_isar_feature(aa32_fhm, s)) { - return 1; - } - if (q) { - rm = vm20; - index = m * 2 + vm3; - } else { - rm = vm20 * 2 + m; - index = vm3; - } - is_long = true; - data = (index << 2) | is_s; /* is_2 == 0 */ - fn_gvec_ptr = gen_helper_gvec_fmlal_idx_a32; - ptr_is_env = true; - } else { - return 1; - } - - VFP_DREG_D(rd, insn); - if (rd & q) { - return 1; - } - if (q || !is_long) { - VFP_DREG_N(rn, insn); - if (rn & q & !is_long) { - return 1; - } - off_rn = vfp_reg_offset(1, rn); - off_rm = vfp_reg_offset(1, rm); - } else { - rn = VFP_SREG_N(insn); - off_rn = vfp_reg_offset(0, rn); - off_rm = vfp_reg_offset(0, rm); - } - if (s->fp_excp_el) { - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, - syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); - return 0; - } - if (!s->vfp_enabled) { - return 1; - } - - opr_sz = (1 + q) * 8; - if (fn_gvec_ptr) { - TCGv_ptr ptr; - if (ptr_is_env) { - ptr = cpu_env; - } else { - ptr = get_fpstatus_ptr(1); - } - tcg_gen_gvec_3_ptr(vfp_reg_offset(1, rd), off_rn, off_rm, ptr, - opr_sz, opr_sz, data, fn_gvec_ptr); - if (!ptr_is_env) { - tcg_temp_free_ptr(ptr); - } - } else { - tcg_gen_gvec_3_ool(vfp_reg_offset(1, rd), off_rn, off_rm, - opr_sz, opr_sz, data, fn_gvec); - } - return 0; -} - static int disas_coproc_insn(DisasContext *s, uint32_t insn) { int cpnum, is64, crn, crm, opc1, opc2, isread, rt, rt2; @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } } } - } else if ((insn & 0x0f000a00) == 0x0e000800 - && arm_dc_feature(s, ARM_FEATURE_V8)) { - if (disas_neon_insn_2reg_scalar_ext(s, insn)) { - goto illegal_op; - } - return; } goto illegal_op; } @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } break; } - if ((insn & 0xff000a00) == 0xfe000800 - && arm_dc_feature(s, ARM_FEATURE_V8)) { - /* The Thumb2 and ARM encodings are identical. */ - if (disas_neon_insn_2reg_scalar_ext(s, insn)) { - goto illegal_op; - } - } else if (((insn >> 24) & 3) == 3) { + if (((insn >> 24) & 3) == 3) { /* Translate into the equivalent ARM encoding. */ insn = (insn & 0xe2ffffff) | ((insn & (1 << 28)) >> 4) | (1 << 28); if (disas_neon_data_insn(s, insn)) { -- 2.20.1 Convert the Neon "load/store multiple structures" insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-12-peter.maydell@linaro.org --- target/arm/neon-ls.decode | 7 ++ target/arm/translate-neon.inc.c | 124 ++++++++++++++++++++++++++++++++ target/arm/translate.c | 91 +---------------------- 3 files changed, 133 insertions(+), 89 deletions(-) diff --git a/target/arm/neon-ls.decode b/target/arm/neon-ls.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-ls.decode +++ b/target/arm/neon-ls.decode @@ -XXX,XX +XXX,XX @@ # 0b1111_1001_xxx0_xxxx_xxxx_xxxx_xxxx_xxxx # This file works on the A32 encoding only; calling code for T32 has to # transform the insn into the A32 version first. + +%vd_dp 22:1 12:4 + +# Neon load/store multiple structures + +VLDST_multiple 1111 0100 0 . l:1 0 rn:4 .... itype:4 size:2 align:2 rm:4 \ + vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VFML_scalar(DisasContext *s, arg_VFML_scalar *a) gen_helper_gvec_fmlal_idx_a32); return true; } + +static struct { + int nregs; + int interleave; + int spacing; +} const neon_ls_element_type[11] = { + {1, 4, 1}, + {1, 4, 2}, + {4, 1, 1}, + {2, 2, 2}, + {1, 3, 1}, + {1, 3, 2}, + {3, 1, 1}, + {1, 1, 1}, + {1, 2, 1}, + {1, 2, 2}, + {2, 1, 1} +}; + +static void gen_neon_ldst_base_update(DisasContext *s, int rm, int rn, + int stride) +{ + if (rm != 15) { + TCGv_i32 base; + + base = load_reg(s, rn); + if (rm == 13) { + tcg_gen_addi_i32(base, base, stride); + } else { + TCGv_i32 index; + index = load_reg(s, rm); + tcg_gen_add_i32(base, base, index); + tcg_temp_free_i32(index); + } + store_reg(s, rn, base); + } +} + +static bool trans_VLDST_multiple(DisasContext *s, arg_VLDST_multiple *a) +{ + /* Neon load/store multiple structures */ + int nregs, interleave, spacing, reg, n; + MemOp endian = s->be_data; + int mmu_idx = get_mem_index(s); + int size = a->size; + TCGv_i64 tmp64; + TCGv_i32 addr, tmp; + + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist */ + if (!dc_isar_feature(aa32_simd_r32, s) && (a->vd & 0x10)) { + return false; + } + if (a->itype > 10) { + return false; + } + /* Catch UNDEF cases for bad values of align field */ + switch (a->itype & 0xc) { + case 4: + if (a->align >= 2) { + return false; + } + break; + case 8: + if (a->align == 3) { + return false; + } + break; + default: + break; + } + nregs = neon_ls_element_type[a->itype].nregs; + interleave = neon_ls_element_type[a->itype].interleave; + spacing = neon_ls_element_type[a->itype].spacing; + if (size == 3 && (interleave | spacing) != 1) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* For our purposes, bytes are always little-endian. */ + if (size == 0) { + endian = MO_LE; + } + /* + * Consecutive little-endian elements from a single register + * can be promoted to a larger little-endian operation. + */ + if (interleave == 1 && endian == MO_LE) { + size = 3; + } + tmp64 = tcg_temp_new_i64(); + addr = tcg_temp_new_i32(); + tmp = tcg_const_i32(1 << size); + load_reg_var(s, addr, a->rn); + for (reg = 0; reg < nregs; reg++) { + for (n = 0; n < 8 >> size; n++) { + int xs; + for (xs = 0; xs < interleave; xs++) { + int tt = a->vd + reg + spacing * xs; + + if (a->l) { + gen_aa32_ld_i64(s, tmp64, addr, mmu_idx, endian | size); + neon_store_element64(tt, n, size, tmp64); + } else { + neon_load_element64(tmp64, tt, n, size); + gen_aa32_st_i64(s, tmp64, addr, mmu_idx, endian | size); + } + tcg_gen_add_i32(addr, addr, tmp); + } + } + } + tcg_temp_free_i32(addr); + tcg_temp_free_i32(tmp); + tcg_temp_free_i64(tmp64); + + gen_neon_ldst_base_update(s, a->rm, a->rn, nregs * interleave * 8); + return true; +} 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 gen_neon_trn_u16(TCGv_i32 t0, TCGv_i32 t1) } -static struct { - int nregs; - int interleave; - int spacing; -} const neon_ls_element_type[11] = { - {1, 4, 1}, - {1, 4, 2}, - {4, 1, 1}, - {2, 2, 2}, - {1, 3, 1}, - {1, 3, 2}, - {3, 1, 1}, - {1, 1, 1}, - {1, 2, 1}, - {1, 2, 2}, - {2, 1, 1} -}; - /* Translate a NEON load/store element instruction. Return nonzero if the instruction is invalid. */ static int disas_neon_ls_insn(DisasContext *s, uint32_t insn) { int rd, rn, rm; - int op; int nregs; - int interleave; - int spacing; int stride; int size; int reg; int load; - int n; int vec_size; - int mmu_idx; - MemOp endian; TCGv_i32 addr; TCGv_i32 tmp; - TCGv_i32 tmp2; - TCGv_i64 tmp64; if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { return 1; @@ -XXX,XX +XXX,XX @@ static int disas_neon_ls_insn(DisasContext *s, uint32_t insn) rn = (insn >> 16) & 0xf; rm = insn & 0xf; load = (insn & (1 << 21)) != 0; - endian = s->be_data; - mmu_idx = get_mem_index(s); if ((insn & (1 << 23)) == 0) { - /* Load store all elements. */ - op = (insn >> 8) & 0xf; - size = (insn >> 6) & 3; - if (op > 10) - return 1; - /* Catch UNDEF cases for bad values of align field */ - switch (op & 0xc) { - case 4: - if (((insn >> 5) & 1) == 1) { - return 1; - } - break; - case 8: - if (((insn >> 4) & 3) == 3) { - return 1; - } - break; - default: - break; - } - nregs = neon_ls_element_type[op].nregs; - interleave = neon_ls_element_type[op].interleave; - spacing = neon_ls_element_type[op].spacing; - if (size == 3 && (interleave | spacing) != 1) { - return 1; - } - /* For our purposes, bytes are always little-endian. */ - if (size == 0) { - endian = MO_LE; - } - /* Consecutive little-endian elements from a single register - * can be promoted to a larger little-endian operation. - */ - if (interleave == 1 && endian == MO_LE) { - size = 3; - } - tmp64 = tcg_temp_new_i64(); - addr = tcg_temp_new_i32(); - tmp2 = tcg_const_i32(1 << size); - load_reg_var(s, addr, rn); - for (reg = 0; reg < nregs; reg++) { - for (n = 0; n < 8 >> size; n++) { - int xs; - for (xs = 0; xs < interleave; xs++) { - int tt = rd + reg + spacing * xs; - - if (load) { - gen_aa32_ld_i64(s, tmp64, addr, mmu_idx, endian | size); - neon_store_element64(tt, n, size, tmp64); - } else { - neon_load_element64(tmp64, tt, n, size); - gen_aa32_st_i64(s, tmp64, addr, mmu_idx, endian | size); - } - tcg_gen_add_i32(addr, addr, tmp2); - } - } - } - tcg_temp_free_i32(addr); - tcg_temp_free_i32(tmp2); - tcg_temp_free_i64(tmp64); - stride = nregs * interleave * 8; + /* Load store all elements -- handled already by decodetree */ + return 1; } else { size = (insn >> 10) & 3; if (size == 3) { -- 2.20.1 Convert the Neon "load single structure to all lanes" insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-13-peter.maydell@linaro.org --- target/arm/neon-ls.decode | 5 +++ target/arm/translate-neon.inc.c | 73 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 55 +------------------------ 3 files changed, 80 insertions(+), 53 deletions(-) diff --git a/target/arm/neon-ls.decode b/target/arm/neon-ls.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-ls.decode +++ b/target/arm/neon-ls.decode @@ -XXX,XX +XXX,XX @@ VLDST_multiple 1111 0100 0 . l:1 0 rn:4 .... itype:4 size:2 align:2 rm:4 \ vd=%vd_dp + +# Neon load single element to all lanes + +VLD_all_lanes 1111 0100 1 . 1 0 rn:4 .... 11 n:2 size:2 t:1 a:1 rm:4 \ + vd=%vd_dp diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VLDST_multiple(DisasContext *s, arg_VLDST_multiple *a) gen_neon_ldst_base_update(s, a->rm, a->rn, nregs * interleave * 8); return true; } + +static bool trans_VLD_all_lanes(DisasContext *s, arg_VLD_all_lanes *a) +{ + /* Neon load single structure to all lanes */ + int reg, stride, vec_size; + int vd = a->vd; + int size = a->size; + int nregs = a->n + 1; + TCGv_i32 addr, tmp; + + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist */ + if (!dc_isar_feature(aa32_simd_r32, s) && (a->vd & 0x10)) { + return false; + } + + if (size == 3) { + if (nregs != 4 || a->a == 0) { + return false; + } + /* For VLD4 size == 3 a == 1 means 32 bits at 16 byte alignment */ + size = 2; + } + if (nregs == 1 && a->a == 1 && size == 0) { + return false; + } + if (nregs == 3 && a->a == 1) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + /* + * VLD1 to all lanes: T bit indicates how many Dregs to write. + * VLD2/3/4 to all lanes: T bit indicates register stride. + */ + stride = a->t ? 2 : 1; + vec_size = nregs == 1 ? stride * 8 : 8; + + tmp = tcg_temp_new_i32(); + addr = tcg_temp_new_i32(); + load_reg_var(s, addr, a->rn); + for (reg = 0; reg < nregs; reg++) { + gen_aa32_ld_i32(s, tmp, addr, get_mem_index(s), + s->be_data | size); + if ((vd & 1) && vec_size == 16) { + /* + * We cannot write 16 bytes at once because the + * destination is unaligned. + */ + tcg_gen_gvec_dup_i32(size, neon_reg_offset(vd, 0), + 8, 8, tmp); + tcg_gen_gvec_mov(0, neon_reg_offset(vd + 1, 0), + neon_reg_offset(vd, 0), 8, 8); + } else { + tcg_gen_gvec_dup_i32(size, neon_reg_offset(vd, 0), + vec_size, vec_size, tmp); + } + tcg_gen_addi_i32(addr, addr, 1 << size); + vd += stride; + } + tcg_temp_free_i32(tmp); + tcg_temp_free_i32(addr); + + gen_neon_ldst_base_update(s, a->rm, a->rn, (1 << size) * nregs); + + return true; +} 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 int disas_neon_ls_insn(DisasContext *s, uint32_t insn) int size; int reg; int load; - int vec_size; TCGv_i32 addr; TCGv_i32 tmp; @@ -XXX,XX +XXX,XX @@ static int disas_neon_ls_insn(DisasContext *s, uint32_t insn) } else { size = (insn >> 10) & 3; if (size == 3) { - /* Load single element to all lanes. */ - int a = (insn >> 4) & 1; - if (!load) { - return 1; - } - size = (insn >> 6) & 3; - nregs = ((insn >> 8) & 3) + 1; - - if (size == 3) { - if (nregs != 4 || a == 0) { - return 1; - } - /* For VLD4 size==3 a == 1 means 32 bits at 16 byte alignment */ - size = 2; - } - if (nregs == 1 && a == 1 && size == 0) { - return 1; - } - if (nregs == 3 && a == 1) { - return 1; - } - addr = tcg_temp_new_i32(); - load_reg_var(s, addr, rn); - - /* VLD1 to all lanes: bit 5 indicates how many Dregs to write. - * VLD2/3/4 to all lanes: bit 5 indicates register stride. - */ - stride = (insn & (1 << 5)) ? 2 : 1; - vec_size = nregs == 1 ? stride * 8 : 8; - - tmp = tcg_temp_new_i32(); - for (reg = 0; reg < nregs; reg++) { - gen_aa32_ld_i32(s, tmp, addr, get_mem_index(s), - s->be_data | size); - if ((rd & 1) && vec_size == 16) { - /* We cannot write 16 bytes at once because the - * destination is unaligned. - */ - tcg_gen_gvec_dup_i32(size, neon_reg_offset(rd, 0), - 8, 8, tmp); - tcg_gen_gvec_mov(0, neon_reg_offset(rd + 1, 0), - neon_reg_offset(rd, 0), 8, 8); - } else { - tcg_gen_gvec_dup_i32(size, neon_reg_offset(rd, 0), - vec_size, vec_size, tmp); - } - tcg_gen_addi_i32(addr, addr, 1 << size); - rd += stride; - } - tcg_temp_free_i32(tmp); - tcg_temp_free_i32(addr); - stride = (1 << size) * nregs; + /* Load single element to all lanes -- handled by decodetree */ + return 1; } else { /* Single element. */ int idx = (insn >> 4) & 0xf; -- 2.20.1 Convert the Neon "load/store single structure to one lane" insns to decodetree. As this is the last set of insns in the neon load/store group, we can remove the whole disas_neon_ls_insn() function. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-14-peter.maydell@linaro.org --- target/arm/neon-ls.decode | 11 +++ target/arm/translate-neon.inc.c | 89 +++++++++++++++++++ target/arm/translate.c | 147 -------------------------------- 3 files changed, 100 insertions(+), 147 deletions(-) diff --git a/target/arm/neon-ls.decode b/target/arm/neon-ls.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-ls.decode +++ b/target/arm/neon-ls.decode @@ -XXX,XX +XXX,XX @@ VLDST_multiple 1111 0100 0 . l:1 0 rn:4 .... itype:4 size:2 align:2 rm:4 \ VLD_all_lanes 1111 0100 1 . 1 0 rn:4 .... 11 n:2 size:2 t:1 a:1 rm:4 \ vd=%vd_dp + +# Neon load/store single structure to one lane +%imm1_5_p1 5:1 !function=plus1 +%imm1_6_p1 6:1 !function=plus1 + +VLDST_single 1111 0100 1 . l:1 0 rn:4 .... 00 n:2 reg_idx:3 align:1 rm:4 \ + vd=%vd_dp size=0 stride=1 +VLDST_single 1111 0100 1 . l:1 0 rn:4 .... 01 n:2 reg_idx:2 align:2 rm:4 \ + vd=%vd_dp size=1 stride=%imm1_5_p1 +VLDST_single 1111 0100 1 . l:1 0 rn:4 .... 10 n:2 reg_idx:1 align:3 rm:4 \ + vd=%vd_dp size=2 stride=%imm1_6_p1 diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ * It might be possible to convert it to a standalone .c file eventually. */ +static inline int plus1(DisasContext *s, int x) +{ + return x + 1; +} + /* Include the generated Neon decoder */ #include "decode-neon-dp.inc.c" #include "decode-neon-ls.inc.c" @@ -XXX,XX +XXX,XX @@ static bool trans_VLD_all_lanes(DisasContext *s, arg_VLD_all_lanes *a) return true; } + +static bool trans_VLDST_single(DisasContext *s, arg_VLDST_single *a) +{ + /* Neon load/store single structure to one lane */ + int reg; + int nregs = a->n + 1; + int vd = a->vd; + TCGv_i32 addr, tmp; + + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist */ + if (!dc_isar_feature(aa32_simd_r32, s) && (a->vd & 0x10)) { + return false; + } + + /* Catch the UNDEF cases. This is unavoidably a bit messy. */ + switch (nregs) { + case 1: + if (((a->align & (1 << a->size)) != 0) || + (a->size == 2 && ((a->align & 3) == 1 || (a->align & 3) == 2))) { + return false; + } + break; + case 3: + if ((a->align & 1) != 0) { + return false; + } + /* fall through */ + case 2: + if (a->size == 2 && (a->align & 2) != 0) { + return false; + } + break; + case 4: + if ((a->size == 2) && ((a->align & 3) == 3)) { + return false; + } + break; + default: + abort(); + } + if ((vd + a->stride * (nregs - 1)) > 31) { + /* + * Attempts to write off the end of the register file are + * UNPREDICTABLE; we choose to UNDEF because otherwise we would + * access off the end of the array that holds the register data. + */ + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + tmp = tcg_temp_new_i32(); + addr = tcg_temp_new_i32(); + load_reg_var(s, addr, a->rn); + /* + * TODO: if we implemented alignment exceptions, we should check + * addr against the alignment encoded in a->align here. + */ + for (reg = 0; reg < nregs; reg++) { + if (a->l) { + gen_aa32_ld_i32(s, tmp, addr, get_mem_index(s), + s->be_data | a->size); + neon_store_element(vd, a->reg_idx, a->size, tmp); + } else { /* Store */ + neon_load_element(tmp, vd, a->reg_idx, a->size); + gen_aa32_st_i32(s, tmp, addr, get_mem_index(s), + s->be_data | a->size); + } + vd += a->stride; + tcg_gen_addi_i32(addr, addr, 1 << a->size); + } + tcg_temp_free_i32(addr); + tcg_temp_free_i32(tmp); + + gen_neon_ldst_base_update(s, a->rm, a->rn, (1 << a->size) * nregs); + + return true; +} 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 gen_neon_trn_u16(TCGv_i32 t0, TCGv_i32 t1) tcg_temp_free_i32(rd); } - -/* Translate a NEON load/store element instruction. Return nonzero if the - instruction is invalid. */ -static int disas_neon_ls_insn(DisasContext *s, uint32_t insn) -{ - int rd, rn, rm; - int nregs; - int stride; - int size; - int reg; - int load; - TCGv_i32 addr; - TCGv_i32 tmp; - - if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { - return 1; - } - - /* FIXME: this access check should not take precedence over UNDEF - * for invalid encodings; we will generate incorrect syndrome information - * for attempts to execute invalid vfp/neon encodings with FP disabled. - */ - if (s->fp_excp_el) { - gen_exception_insn(s, s->pc_curr, EXCP_UDEF, - syn_simd_access_trap(1, 0xe, false), s->fp_excp_el); - return 0; - } - - if (!s->vfp_enabled) - return 1; - VFP_DREG_D(rd, insn); - rn = (insn >> 16) & 0xf; - rm = insn & 0xf; - load = (insn & (1 << 21)) != 0; - if ((insn & (1 << 23)) == 0) { - /* Load store all elements -- handled already by decodetree */ - return 1; - } else { - size = (insn >> 10) & 3; - if (size == 3) { - /* Load single element to all lanes -- handled by decodetree */ - return 1; - } else { - /* Single element. */ - int idx = (insn >> 4) & 0xf; - int reg_idx; - switch (size) { - case 0: - reg_idx = (insn >> 5) & 7; - stride = 1; - break; - case 1: - reg_idx = (insn >> 6) & 3; - stride = (insn & (1 << 5)) ? 2 : 1; - break; - case 2: - reg_idx = (insn >> 7) & 1; - stride = (insn & (1 << 6)) ? 2 : 1; - break; - default: - abort(); - } - nregs = ((insn >> 8) & 3) + 1; - /* Catch the UNDEF cases. This is unavoidably a bit messy. */ - switch (nregs) { - case 1: - if (((idx & (1 << size)) != 0) || - (size == 2 && ((idx & 3) == 1 || (idx & 3) == 2))) { - return 1; - } - break; - case 3: - if ((idx & 1) != 0) { - return 1; - } - /* fall through */ - case 2: - if (size == 2 && (idx & 2) != 0) { - return 1; - } - break; - case 4: - if ((size == 2) && ((idx & 3) == 3)) { - return 1; - } - break; - default: - abort(); - } - if ((rd + stride * (nregs - 1)) > 31) { - /* Attempts to write off the end of the register file - * are UNPREDICTABLE; we choose to UNDEF because otherwise - * the neon_load_reg() would write off the end of the array. - */ - return 1; - } - tmp = tcg_temp_new_i32(); - addr = tcg_temp_new_i32(); - load_reg_var(s, addr, rn); - for (reg = 0; reg < nregs; reg++) { - if (load) { - gen_aa32_ld_i32(s, tmp, addr, get_mem_index(s), - s->be_data | size); - neon_store_element(rd, reg_idx, size, tmp); - } else { /* Store */ - neon_load_element(tmp, rd, reg_idx, size); - gen_aa32_st_i32(s, tmp, addr, get_mem_index(s), - s->be_data | size); - } - rd += stride; - tcg_gen_addi_i32(addr, addr, 1 << size); - } - tcg_temp_free_i32(addr); - tcg_temp_free_i32(tmp); - stride = nregs * (1 << size); - } - } - if (rm != 15) { - TCGv_i32 base; - - base = load_reg(s, rn); - if (rm == 13) { - tcg_gen_addi_i32(base, base, stride); - } else { - TCGv_i32 index; - index = load_reg(s, rm); - tcg_gen_add_i32(base, base, index); - tcg_temp_free_i32(index); - } - store_reg(s, rn, base); - } - return 0; -} - static inline void gen_neon_narrow(int size, TCGv_i32 dest, TCGv_i64 src) { switch (size) { @@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn) } return; } - if ((insn & 0x0f100000) == 0x04000000) { - /* NEON load/store. */ - if (disas_neon_ls_insn(s, insn)) { - goto illegal_op; - } - return; - } if ((insn & 0x0e000f00) == 0x0c000100) { if (arm_dc_feature(s, ARM_FEATURE_IWMMXT)) { /* iWMMXt register transfer. */ @@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn) } break; case 12: - if ((insn & 0x01100000) == 0x01000000) { - if (disas_neon_ls_insn(s, insn)) { - goto illegal_op; - } - break; - } goto illegal_op; default: illegal_op: -- 2.20.1 Convert the Neon 3-reg-same VADD and VSUB insns to decodetree. Note that we don't need the neon_3r_sizes[op] check here because all size values are OK for VADD and VSUB; we'll add this when we convert the first insn that has size restrictions. For this we need one of the GVecGen*Fn typedefs currently in translate-a64.h; move them all to translate.h as a block so they are visible to the 32-bit decoder. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-15-peter.maydell@linaro.org --- target/arm/translate-a64.h | 9 -------- target/arm/translate.h | 9 ++++++++ target/arm/neon-dp.decode | 17 +++++++++++++++ target/arm/translate-neon.inc.c | 38 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 14 ++++-------- 5 files changed, 68 insertions(+), 19 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 @@ static inline int vec_full_reg_size(DisasContext *s) bool disas_sve(DisasContext *, uint32_t); -/* Note that the gvec expanders operate on offsets + sizes. */ -typedef void GVecGen2Fn(unsigned, uint32_t, uint32_t, uint32_t, uint32_t); -typedef void GVecGen2iFn(unsigned, uint32_t, uint32_t, int64_t, - uint32_t, uint32_t); -typedef void GVecGen3Fn(unsigned, uint32_t, uint32_t, - uint32_t, uint32_t, uint32_t); -typedef void GVecGen4Fn(unsigned, uint32_t, uint32_t, uint32_t, - uint32_t, uint32_t, uint32_t); - #endif /* TARGET_ARM_TRANSLATE_A64_H */ 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 @@ void gen_sshl_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b); #define dc_isar_feature(name, ctx) \ ({ DisasContext *ctx_ = (ctx); isar_feature_##name(ctx_->isar); }) +/* Note that the gvec expanders operate on offsets + sizes. */ +typedef void GVecGen2Fn(unsigned, uint32_t, uint32_t, uint32_t, uint32_t); +typedef void GVecGen2iFn(unsigned, uint32_t, uint32_t, int64_t, + uint32_t, uint32_t); +typedef void GVecGen3Fn(unsigned, uint32_t, uint32_t, + uint32_t, uint32_t, uint32_t); +typedef void GVecGen4Fn(unsigned, uint32_t, uint32_t, uint32_t, + uint32_t, uint32_t, uint32_t); + #endif /* TARGET_ARM_TRANSLATE_H */ diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ # # This file is processed by scripts/decodetree.py # +# VFP/Neon register fields; same as vfp.decode +%vm_dp 5:1 0:4 +%vn_dp 7:1 16:4 +%vd_dp 22:1 12:4 # Encodings for Neon data processing instructions where the T32 encoding # is a simple transformation of the A32 encoding. @@ -XXX,XX +XXX,XX @@ # 0b111p_1111_qqqq_qqqq_qqqq_qqqq_qqqq_qqqq # This file works on the A32 encoding only; calling code for T32 has to # transform the insn into the A32 version first. + +###################################################################### +# 3-reg-same grouping: +# 1111 001 U 0 D sz:2 Vn:4 Vd:4 opc:4 N Q M op Vm:4 +###################################################################### + +&3same vm vn vd q size + +@3same .... ... . . . size:2 .... .... .... . q:1 . . .... \ + &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp + +VADD_3s 1111 001 0 0 . .. .... .... 1000 . . . 0 .... @3same +VSUB_3s 1111 001 1 0 . .. .... .... 1000 . . . 0 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool trans_VLDST_single(DisasContext *s, arg_VLDST_single *a) return true; } + +static bool do_3same(DisasContext *s, arg_3same *a, GVecGen3Fn fn) +{ + int vec_size = a->q ? 16 : 8; + int rd_ofs = neon_reg_offset(a->vd, 0); + int rn_ofs = neon_reg_offset(a->vn, 0); + int rm_ofs = neon_reg_offset(a->vm, 0); + + if (!arm_dc_feature(s, ARM_FEATURE_NEON)) { + return false; + } + + /* UNDEF accesses to D16-D31 if they don't exist. */ + if (!dc_isar_feature(aa32_simd_r32, s) && + ((a->vd | a->vn | a->vm) & 0x10)) { + return false; + } + + if ((a->vn | a->vm | a->vd) & a->q) { + return false; + } + + if (!vfp_access_check(s)) { + return true; + } + + fn(a->size, rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size); + return true; +} + +#define DO_3SAME(INSN, FUNC) \ + static bool trans_##INSN##_3s(DisasContext *s, arg_3same *a) \ + { \ + return do_3same(s, a, FUNC); \ + } + +DO_3SAME(VADD, tcg_gen_gvec_add) +DO_3SAME(VSUB, tcg_gen_gvec_sub) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) } return 0; - case NEON_3R_VADD_VSUB: - if (u) { - tcg_gen_gvec_sub(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } else { - tcg_gen_gvec_add(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } - return 0; - case NEON_3R_VQADD: tcg_gen_gvec_4(rd_ofs, offsetof(CPUARMState, vfp.qc), rn_ofs, rm_ofs, vec_size, vec_size, @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) tcg_gen_gvec_3(rd_ofs, rm_ofs, rn_ofs, vec_size, vec_size, u ? &ushl_op[size] : &sshl_op[size]); return 0; + + case NEON_3R_VADD_VSUB: + /* Already handled by decodetree */ + return 1; } if (size == 3) { -- 2.20.1 Convert the Neon logic ops in the 3-reg-same grouping to decodetree. Note that for the logic ops the 'size' field forms part of their decode and the actual operations are always bitwise. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-16-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 12 +++++++++++ target/arm/translate-neon.inc.c | 19 +++++++++++++++++ target/arm/translate.c | 38 +-------------------------------- 3 files changed, 32 insertions(+), 37 deletions(-) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ @3same .... ... . . . size:2 .... .... .... . q:1 . . .... \ &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp +@3same_logic .... ... . . . .. .... .... .... . q:1 .. .... \ + &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp size=0 + +VAND_3s 1111 001 0 0 . 00 .... .... 0001 ... 1 .... @3same_logic +VBIC_3s 1111 001 0 0 . 01 .... .... 0001 ... 1 .... @3same_logic +VORR_3s 1111 001 0 0 . 10 .... .... 0001 ... 1 .... @3same_logic +VORN_3s 1111 001 0 0 . 11 .... .... 0001 ... 1 .... @3same_logic +VEOR_3s 1111 001 1 0 . 00 .... .... 0001 ... 1 .... @3same_logic +VBSL_3s 1111 001 1 0 . 01 .... .... 0001 ... 1 .... @3same_logic +VBIT_3s 1111 001 1 0 . 10 .... .... 0001 ... 1 .... @3same_logic +VBIF_3s 1111 001 1 0 . 11 .... .... 0001 ... 1 .... @3same_logic + VADD_3s 1111 001 0 0 . .. .... .... 1000 . . . 0 .... @3same VSUB_3s 1111 001 1 0 . .. .... .... 1000 . . . 0 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static bool do_3same(DisasContext *s, arg_3same *a, GVecGen3Fn fn) DO_3SAME(VADD, tcg_gen_gvec_add) DO_3SAME(VSUB, tcg_gen_gvec_sub) +DO_3SAME(VAND, tcg_gen_gvec_and) +DO_3SAME(VBIC, tcg_gen_gvec_andc) +DO_3SAME(VORR, tcg_gen_gvec_or) +DO_3SAME(VORN, tcg_gen_gvec_orc) +DO_3SAME(VEOR, tcg_gen_gvec_xor) + +/* These insns are all gvec_bitsel but with the inputs in various orders. */ +#define DO_3SAME_BITSEL(INSN, O1, O2, O3) \ + static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + tcg_gen_gvec_bitsel(vece, rd_ofs, O1, O2, O3, oprsz, maxsz); \ + } \ + DO_3SAME(INSN, gen_##INSN##_3s) + +DO_3SAME_BITSEL(VBSL, rd_ofs, rn_ofs, rm_ofs) +DO_3SAME_BITSEL(VBIT, rm_ofs, rn_ofs, rd_ofs) +DO_3SAME_BITSEL(VBIF, rm_ofs, rd_ofs, rn_ofs) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) } return 1; - case NEON_3R_LOGIC: /* Logic ops. */ - switch ((u << 2) | size) { - case 0: /* VAND */ - tcg_gen_gvec_and(0, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 1: /* VBIC */ - tcg_gen_gvec_andc(0, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 2: /* VORR */ - tcg_gen_gvec_or(0, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 3: /* VORN */ - tcg_gen_gvec_orc(0, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 4: /* VEOR */ - tcg_gen_gvec_xor(0, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 5: /* VBSL */ - tcg_gen_gvec_bitsel(MO_8, rd_ofs, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - break; - case 6: /* VBIT */ - tcg_gen_gvec_bitsel(MO_8, rd_ofs, rm_ofs, rn_ofs, rd_ofs, - vec_size, vec_size); - break; - case 7: /* VBIF */ - tcg_gen_gvec_bitsel(MO_8, rd_ofs, rm_ofs, rd_ofs, rn_ofs, - vec_size, vec_size); - break; - } - return 0; - case NEON_3R_VQADD: tcg_gen_gvec_4(rd_ofs, offsetof(CPUARMState, vfp.qc), rn_ofs, rm_ofs, vec_size, vec_size, @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) return 0; case NEON_3R_VADD_VSUB: + case NEON_3R_LOGIC: /* Already handled by decodetree */ return 1; } -- 2.20.1 Convert the Neon 3-reg-same VMAX and VMIN insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-17-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 5 +++++ target/arm/translate-neon.inc.c | 14 ++++++++++++++ target/arm/translate.c | 21 ++------------------- 3 files changed, 21 insertions(+), 19 deletions(-) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ VBSL_3s 1111 001 1 0 . 01 .... .... 0001 ... 1 .... @3same_logic VBIT_3s 1111 001 1 0 . 10 .... .... 0001 ... 1 .... @3same_logic VBIF_3s 1111 001 1 0 . 11 .... .... 0001 ... 1 .... @3same_logic +VMAX_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 0 .... @3same +VMAX_U_3s 1111 001 1 0 . .. .... .... 0110 . . . 0 .... @3same +VMIN_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 1 .... @3same +VMIN_U_3s 1111 001 1 0 . .. .... .... 0110 . . . 1 .... @3same + VADD_3s 1111 001 0 0 . .. .... .... 1000 . . . 0 .... @3same VSUB_3s 1111 001 1 0 . .. .... .... 1000 . . . 0 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ DO_3SAME(VEOR, tcg_gen_gvec_xor) DO_3SAME_BITSEL(VBSL, rd_ofs, rn_ofs, rm_ofs) DO_3SAME_BITSEL(VBIT, rm_ofs, rn_ofs, rd_ofs) DO_3SAME_BITSEL(VBIF, rm_ofs, rd_ofs, rn_ofs) + +#define DO_3SAME_NO_SZ_3(INSN, FUNC) \ + static bool trans_##INSN##_3s(DisasContext *s, arg_3same *a) \ + { \ + if (a->size == 3) { \ + return false; \ + } \ + return do_3same(s, a, FUNC); \ + } + +DO_3SAME_NO_SZ_3(VMAX_S, tcg_gen_gvec_smax) +DO_3SAME_NO_SZ_3(VMAX_U, tcg_gen_gvec_umax) +DO_3SAME_NO_SZ_3(VMIN_S, tcg_gen_gvec_smin) +DO_3SAME_NO_SZ_3(VMIN_U, tcg_gen_gvec_umin) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size); return 0; - case NEON_3R_VMAX: - if (u) { - tcg_gen_gvec_umax(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } else { - tcg_gen_gvec_smax(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } - return 0; - case NEON_3R_VMIN: - if (u) { - tcg_gen_gvec_umin(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } else { - tcg_gen_gvec_smin(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } - return 0; - case NEON_3R_VSHL: /* Note the operation is vshl vd,vm,vn */ tcg_gen_gvec_3(rd_ofs, rm_ofs, rn_ofs, vec_size, vec_size, @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) case NEON_3R_VADD_VSUB: case NEON_3R_LOGIC: + case NEON_3R_VMAX: + case NEON_3R_VMIN: /* Already handled by decodetree */ return 1; } -- 2.20.1 Convert the Neon comparison ops in the 3-reg-same grouping to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-18-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 8 ++++++++ target/arm/translate-neon.inc.c | 22 ++++++++++++++++++++++ target/arm/translate.c | 23 +++-------------------- 3 files changed, 33 insertions(+), 20 deletions(-) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ VBSL_3s 1111 001 1 0 . 01 .... .... 0001 ... 1 .... @3same_logic VBIT_3s 1111 001 1 0 . 10 .... .... 0001 ... 1 .... @3same_logic VBIF_3s 1111 001 1 0 . 11 .... .... 0001 ... 1 .... @3same_logic +VCGT_S_3s 1111 001 0 0 . .. .... .... 0011 . . . 0 .... @3same +VCGT_U_3s 1111 001 1 0 . .. .... .... 0011 . . . 0 .... @3same +VCGE_S_3s 1111 001 0 0 . .. .... .... 0011 . . . 1 .... @3same +VCGE_U_3s 1111 001 1 0 . .. .... .... 0011 . . . 1 .... @3same + VMAX_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 0 .... @3same VMAX_U_3s 1111 001 1 0 . .. .... .... 0110 . . . 0 .... @3same VMIN_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 1 .... @3same @@ -XXX,XX +XXX,XX @@ VMIN_U_3s 1111 001 1 0 . .. .... .... 0110 . . . 1 .... @3same VADD_3s 1111 001 0 0 . .. .... .... 1000 . . . 0 .... @3same VSUB_3s 1111 001 1 0 . .. .... .... 1000 . . . 0 .... @3same + +VTST_3s 1111 001 0 0 . .. .... .... 1000 . . . 1 .... @3same +VCEQ_3s 1111 001 1 0 . .. .... .... 1000 . . . 1 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ DO_3SAME_NO_SZ_3(VMAX_S, tcg_gen_gvec_smax) DO_3SAME_NO_SZ_3(VMAX_U, tcg_gen_gvec_umax) DO_3SAME_NO_SZ_3(VMIN_S, tcg_gen_gvec_smin) DO_3SAME_NO_SZ_3(VMIN_U, tcg_gen_gvec_umin) + +#define DO_3SAME_CMP(INSN, COND) \ + static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + tcg_gen_gvec_cmp(COND, vece, rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz); \ + } \ + DO_3SAME_NO_SZ_3(INSN, gen_##INSN##_3s) + +DO_3SAME_CMP(VCGT_S, TCG_COND_GT) +DO_3SAME_CMP(VCGT_U, TCG_COND_GTU) +DO_3SAME_CMP(VCGE_S, TCG_COND_GE) +DO_3SAME_CMP(VCGE_U, TCG_COND_GEU) +DO_3SAME_CMP(VCEQ, TCG_COND_EQ) + +static void gen_VTST_3s(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, + uint32_t rm_ofs, uint32_t oprsz, uint32_t maxsz) +{ + tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz, &cmtst_op[vece]); +} +DO_3SAME_NO_SZ_3(VTST, gen_VTST_3s) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) u ? &mls_op[size] : &mla_op[size]); return 0; - case NEON_3R_VTST_VCEQ: - if (u) { /* VCEQ */ - tcg_gen_gvec_cmp(TCG_COND_EQ, size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } else { /* VTST */ - tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size, &cmtst_op[size]); - } - return 0; - - case NEON_3R_VCGT: - tcg_gen_gvec_cmp(u ? TCG_COND_GTU : TCG_COND_GT, size, - rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size); - return 0; - - case NEON_3R_VCGE: - tcg_gen_gvec_cmp(u ? TCG_COND_GEU : TCG_COND_GE, size, - rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size); - return 0; - case NEON_3R_VSHL: /* Note the operation is vshl vd,vm,vn */ tcg_gen_gvec_3(rd_ofs, rm_ofs, rn_ofs, vec_size, vec_size, @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) case NEON_3R_LOGIC: case NEON_3R_VMAX: case NEON_3R_VMIN: + case NEON_3R_VTST_VCEQ: + case NEON_3R_VCGT: + case NEON_3R_VCGE: /* Already handled by decodetree */ return 1; } -- 2.20.1 Convert the Neon VQADD/VQSUB insns in the 3-reg-same grouping to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-19-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 6 ++++++ target/arm/translate-neon.inc.c | 15 +++++++++++++++ target/arm/translate.c | 14 ++------------ 3 files changed, 23 insertions(+), 12 deletions(-) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ @3same .... ... . . . size:2 .... .... .... . q:1 . . .... \ &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp +VQADD_S_3s 1111 001 0 0 . .. .... .... 0000 . . . 1 .... @3same +VQADD_U_3s 1111 001 1 0 . .. .... .... 0000 . . . 1 .... @3same + @3same_logic .... ... . . . .. .... .... .... . q:1 .. .... \ &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp size=0 @@ -XXX,XX +XXX,XX @@ VBSL_3s 1111 001 1 0 . 01 .... .... 0001 ... 1 .... @3same_logic VBIT_3s 1111 001 1 0 . 10 .... .... 0001 ... 1 .... @3same_logic VBIF_3s 1111 001 1 0 . 11 .... .... 0001 ... 1 .... @3same_logic +VQSUB_S_3s 1111 001 0 0 . .. .... .... 0010 . . . 1 .... @3same +VQSUB_U_3s 1111 001 1 0 . .. .... .... 0010 . . . 1 .... @3same + VCGT_S_3s 1111 001 0 0 . .. .... .... 0011 . . . 0 .... @3same VCGT_U_3s 1111 001 1 0 . .. .... .... 0011 . . . 0 .... @3same VCGE_S_3s 1111 001 0 0 . .. .... .... 0011 . . . 1 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ static void gen_VTST_3s(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz, &cmtst_op[vece]); } DO_3SAME_NO_SZ_3(VTST, gen_VTST_3s) + +#define DO_3SAME_GVEC4(INSN, OPARRAY) \ + static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + tcg_gen_gvec_4(rd_ofs, offsetof(CPUARMState, vfp.qc), \ + rn_ofs, rm_ofs, oprsz, maxsz, &OPARRAY[vece]); \ + } \ + DO_3SAME(INSN, gen_##INSN##_3s) + +DO_3SAME_GVEC4(VQADD_S, sqadd_op) +DO_3SAME_GVEC4(VQADD_U, uqadd_op) +DO_3SAME_GVEC4(VQSUB_S, sqsub_op) +DO_3SAME_GVEC4(VQSUB_U, uqsub_op) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) } return 1; - case NEON_3R_VQADD: - tcg_gen_gvec_4(rd_ofs, offsetof(CPUARMState, vfp.qc), - rn_ofs, rm_ofs, vec_size, vec_size, - (u ? uqadd_op : sqadd_op) + size); - return 0; - - case NEON_3R_VQSUB: - tcg_gen_gvec_4(rd_ofs, offsetof(CPUARMState, vfp.qc), - rn_ofs, rm_ofs, vec_size, vec_size, - (u ? uqsub_op : sqsub_op) + size); - return 0; - case NEON_3R_VMUL: /* VMUL */ if (u) { /* Polynomial case allows only P8. */ @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) case NEON_3R_VTST_VCEQ: case NEON_3R_VCGT: case NEON_3R_VCGE: + case NEON_3R_VQADD: + case NEON_3R_VQSUB: /* Already handled by decodetree */ return 1; } -- 2.20.1 Convert the Neon VMUL, VMLA, VMLS and VSHL insns in the 3-reg-same grouping to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-20-peter.maydell@linaro.org --- target/arm/neon-dp.decode | 9 +++++++ target/arm/translate-neon.inc.c | 44 +++++++++++++++++++++++++++++++++ target/arm/translate.c | 28 +++------------------ 3 files changed, 56 insertions(+), 25 deletions(-) diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -XXX,XX +XXX,XX @@ VCGT_U_3s 1111 001 1 0 . .. .... .... 0011 . . . 0 .... @3same VCGE_S_3s 1111 001 0 0 . .. .... .... 0011 . . . 1 .... @3same VCGE_U_3s 1111 001 1 0 . .. .... .... 0011 . . . 1 .... @3same +VSHL_S_3s 1111 001 0 0 . .. .... .... 0100 . . . 0 .... @3same +VSHL_U_3s 1111 001 1 0 . .. .... .... 0100 . . . 0 .... @3same + VMAX_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 0 .... @3same VMAX_U_3s 1111 001 1 0 . .. .... .... 0110 . . . 0 .... @3same VMIN_S_3s 1111 001 0 0 . .. .... .... 0110 . . . 1 .... @3same @@ -XXX,XX +XXX,XX @@ VSUB_3s 1111 001 1 0 . .. .... .... 1000 . . . 0 .... @3same VTST_3s 1111 001 0 0 . .. .... .... 1000 . . . 1 .... @3same VCEQ_3s 1111 001 1 0 . .. .... .... 1000 . . . 1 .... @3same + +VMLA_3s 1111 001 0 0 . .. .... .... 1001 . . . 0 .... @3same +VMLS_3s 1111 001 1 0 . .. .... .... 1001 . . . 0 .... @3same + +VMUL_3s 1111 001 0 0 . .. .... .... 1001 . . . 1 .... @3same +VMUL_p_3s 1111 001 1 0 . .. .... .... 1001 . . . 1 .... @3same diff --git a/target/arm/translate-neon.inc.c b/target/arm/translate-neon.inc.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/translate-neon.inc.c +++ b/target/arm/translate-neon.inc.c @@ -XXX,XX +XXX,XX @@ DO_3SAME_NO_SZ_3(VMAX_S, tcg_gen_gvec_smax) DO_3SAME_NO_SZ_3(VMAX_U, tcg_gen_gvec_umax) DO_3SAME_NO_SZ_3(VMIN_S, tcg_gen_gvec_smin) DO_3SAME_NO_SZ_3(VMIN_U, tcg_gen_gvec_umin) +DO_3SAME_NO_SZ_3(VMUL, tcg_gen_gvec_mul) #define DO_3SAME_CMP(INSN, COND) \ static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ @@ -XXX,XX +XXX,XX @@ DO_3SAME_GVEC4(VQADD_S, sqadd_op) DO_3SAME_GVEC4(VQADD_U, uqadd_op) DO_3SAME_GVEC4(VQSUB_S, sqsub_op) DO_3SAME_GVEC4(VQSUB_U, uqsub_op) + +static void gen_VMUL_p_3s(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs, + uint32_t rm_ofs, uint32_t oprsz, uint32_t maxsz) +{ + tcg_gen_gvec_3_ool(rd_ofs, rn_ofs, rm_ofs, oprsz, maxsz, + 0, gen_helper_gvec_pmul_b); +} + +static bool trans_VMUL_p_3s(DisasContext *s, arg_3same *a) +{ + if (a->size != 0) { + return false; + } + return do_3same(s, a, gen_VMUL_p_3s); +} + +#define DO_3SAME_GVEC3_NO_SZ_3(INSN, OPARRAY) \ + static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, \ + oprsz, maxsz, &OPARRAY[vece]); \ + } \ + DO_3SAME_NO_SZ_3(INSN, gen_##INSN##_3s) + + +DO_3SAME_GVEC3_NO_SZ_3(VMLA, mla_op) +DO_3SAME_GVEC3_NO_SZ_3(VMLS, mls_op) + +#define DO_3SAME_GVEC3_SHIFT(INSN, OPARRAY) \ + static void gen_##INSN##_3s(unsigned vece, uint32_t rd_ofs, \ + uint32_t rn_ofs, uint32_t rm_ofs, \ + uint32_t oprsz, uint32_t maxsz) \ + { \ + /* Note the operation is vshl vd,vm,vn */ \ + tcg_gen_gvec_3(rd_ofs, rm_ofs, rn_ofs, \ + oprsz, maxsz, &OPARRAY[vece]); \ + } \ + DO_3SAME(INSN, gen_##INSN##_3s) + +DO_3SAME_GVEC3_SHIFT(VSHL_S, sshl_op) +DO_3SAME_GVEC3_SHIFT(VSHL_U, ushl_op) 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 int disas_neon_data_insn(DisasContext *s, uint32_t insn) } return 1; - case NEON_3R_VMUL: /* VMUL */ - if (u) { - /* Polynomial case allows only P8. */ - if (size != 0) { - return 1; - } - tcg_gen_gvec_3_ool(rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size, - 0, gen_helper_gvec_pmul_b); - } else { - tcg_gen_gvec_mul(size, rd_ofs, rn_ofs, rm_ofs, - vec_size, vec_size); - } - return 0; - - case NEON_3R_VML: /* VMLA, VMLS */ - tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, vec_size, vec_size, - u ? &mls_op[size] : &mla_op[size]); - return 0; - - case NEON_3R_VSHL: - /* Note the operation is vshl vd,vm,vn */ - tcg_gen_gvec_3(rd_ofs, rm_ofs, rn_ofs, vec_size, vec_size, - u ? &ushl_op[size] : &sshl_op[size]); - return 0; - case NEON_3R_VADD_VSUB: case NEON_3R_LOGIC: case NEON_3R_VMAX: @@ -XXX,XX +XXX,XX @@ static int disas_neon_data_insn(DisasContext *s, uint32_t insn) case NEON_3R_VCGE: case NEON_3R_VQADD: case NEON_3R_VQSUB: + case NEON_3R_VMUL: + case NEON_3R_VML: + case NEON_3R_VSHL: /* Already handled by decodetree */ return 1; } -- 2.20.1 We're going to want at least some of the NeonGen* typedefs for the refactored 32-bit Neon decoder, so move them all to translate.h since it makes more sense to keep them in one group. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20200430181003.21682-23-peter.maydell@linaro.org --- target/arm/translate.h | 17 +++++++++++++++++ target/arm/translate-a64.c | 17 ----------------- 2 files changed, 17 insertions(+), 17 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 @@ typedef void GVecGen3Fn(unsigned, uint32_t, uint32_t, typedef void GVecGen4Fn(unsigned, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t); +/* Function prototype for gen_ functions for calling Neon helpers */ +typedef void NeonGenOneOpEnvFn(TCGv_i32, TCGv_ptr, TCGv_i32); +typedef void NeonGenTwoOpFn(TCGv_i32, TCGv_i32, TCGv_i32); +typedef void NeonGenTwoOpEnvFn(TCGv_i32, TCGv_ptr, TCGv_i32, TCGv_i32); +typedef void NeonGenTwo64OpFn(TCGv_i64, TCGv_i64, TCGv_i64); +typedef void NeonGenTwo64OpEnvFn(TCGv_i64, TCGv_ptr, TCGv_i64, TCGv_i64); +typedef void NeonGenNarrowFn(TCGv_i32, TCGv_i64); +typedef void NeonGenNarrowEnvFn(TCGv_i32, TCGv_ptr, TCGv_i64); +typedef void NeonGenWidenFn(TCGv_i64, TCGv_i32); +typedef void NeonGenTwoSingleOPFn(TCGv_i32, TCGv_i32, TCGv_i32, TCGv_ptr); +typedef void NeonGenTwoDoubleOPFn(TCGv_i64, TCGv_i64, TCGv_i64, TCGv_ptr); +typedef void NeonGenOneOpFn(TCGv_i64, TCGv_i64); +typedef void CryptoTwoOpFn(TCGv_ptr, TCGv_ptr); +typedef void CryptoThreeOpIntFn(TCGv_ptr, TCGv_ptr, TCGv_i32); +typedef void CryptoThreeOpFn(TCGv_ptr, TCGv_ptr, TCGv_ptr); +typedef void AtomicThreeOpFn(TCGv_i64, TCGv_i64, TCGv_i64, TCGArg, MemOp); + #endif /* TARGET_ARM_TRANSLATE_H */ 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 @@ typedef struct AArch64DecodeTable { AArch64DecodeFn *disas_fn; } AArch64DecodeTable; -/* Function prototype for gen_ functions for calling Neon helpers */ -typedef void NeonGenOneOpEnvFn(TCGv_i32, TCGv_ptr, TCGv_i32); -typedef void NeonGenTwoOpFn(TCGv_i32, TCGv_i32, TCGv_i32); -typedef void NeonGenTwoOpEnvFn(TCGv_i32, TCGv_ptr, TCGv_i32, TCGv_i32); -typedef void NeonGenTwo64OpFn(TCGv_i64, TCGv_i64, TCGv_i64); -typedef void NeonGenTwo64OpEnvFn(TCGv_i64, TCGv_ptr, TCGv_i64, TCGv_i64); -typedef void NeonGenNarrowFn(TCGv_i32, TCGv_i64); -typedef void NeonGenNarrowEnvFn(TCGv_i32, TCGv_ptr, TCGv_i64); -typedef void NeonGenWidenFn(TCGv_i64, TCGv_i32); -typedef void NeonGenTwoSingleOPFn(TCGv_i32, TCGv_i32, TCGv_i32, TCGv_ptr); -typedef void NeonGenTwoDoubleOPFn(TCGv_i64, TCGv_i64, TCGv_i64, TCGv_ptr); -typedef void NeonGenOneOpFn(TCGv_i64, TCGv_i64); -typedef void CryptoTwoOpFn(TCGv_ptr, TCGv_ptr); -typedef void CryptoThreeOpIntFn(TCGv_ptr, TCGv_ptr, TCGv_i32); -typedef void CryptoThreeOpFn(TCGv_ptr, TCGv_ptr, TCGv_ptr); -typedef void AtomicThreeOpFn(TCGv_i64, TCGv_i64, TCGv_i64, TCGArg, MemOp); - /* initialize TCG globals. */ void a64_translate_init(void) { -- 2.20.1 Hi; this mostly contains the first slice of A64 decodetree patches, plus some other minor pieces. It also has the enablement of MTE for KVM guests. thanks -- PMM The following changes since commit d27e7c359330ba7020bdbed7ed2316cb4cf6ffc1: qapi/parser: Drop two bad type hints for now (2023-05-17 10:18:33 -0700) are available in the Git repository at: https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20230518 for you to fetch changes up to 91608e2a44f36e79cb83f863b8a7bb57d2c98061: docs: Convert u2f.txt to rST (2023-05-18 11:40:32 +0100) ---------------------------------------------------------------- target-arm queue: * Fix vd == vm overlap in sve_ldff1_z * Add support for MTE with KVM guests * Add RAZ/WI handling for DBGDTR[TX|RX] * Start of conversion of A64 decoder to decodetree * Saturate L2CTLR_EL1 core count field rather than overflowing * vexpress: Avoid trivial memory leak of 'flashalias' * sbsa-ref: switch default cpu core to Neoverse-N1 * sbsa-ref: use Bochs graphics card instead of VGA * MAINTAINERS: Add Marcin Juszkiewicz to sbsa-ref reviewer list * docs: Convert u2f.txt to rST ---------------------------------------------------------------- Alex Bennée (1): target/arm: add RAZ/WI handling for DBGDTR[TX|RX] Cornelia Huck (1): arm/kvm: add support for MTE Marcin Juszkiewicz (3): sbsa-ref: switch default cpu core to Neoverse-N1 Maintainers: add myself as reviewer for sbsa-ref sbsa-ref: use Bochs graphics card instead of VGA Peter Maydell (14): target/arm: Create decodetree skeleton for A64 target/arm: Pull calls to disas_sve() and disas_sme() out of legacy decoder target/arm: Convert Extract instructions to decodetree target/arm: Convert unconditional branch immediate to decodetree target/arm: Convert CBZ, CBNZ to decodetree target/arm: Convert TBZ, TBNZ to decodetree target/arm: Convert conditional branch insns to decodetree target/arm: Convert BR, BLR, RET to decodetree target/arm: Convert BRA[AB]Z, BLR[AB]Z, RETA[AB] to decodetree target/arm: Convert BRAA, BRAB, BLRAA, BLRAB to decodetree target/arm: Convert ERET, ERETAA, ERETAB to decodetree target/arm: Saturate L2CTLR_EL1 core count field rather than overflowing hw/arm/vexpress: Avoid trivial memory leak of 'flashalias' docs: Convert u2f.txt to rST Richard Henderson (10): target/arm: Fix vd == vm overlap in sve_ldff1_z target/arm: Split out disas_a64_legacy target/arm: Convert PC-rel addressing to decodetree target/arm: Split gen_add_CC and gen_sub_CC target/arm: Convert Add/subtract (immediate) to decodetree target/arm: Convert Add/subtract (immediate with tags) to decodetree target/arm: Replace bitmask64 with MAKE_64BIT_MASK target/arm: Convert Logical (immediate) to decodetree target/arm: Convert Move wide (immediate) to decodetree target/arm: Convert Bitfield to decodetree MAINTAINERS | 1 + docs/system/device-emulation.rst | 1 + docs/system/devices/usb-u2f.rst | 93 +++ docs/system/devices/usb.rst | 2 +- docs/u2f.txt | 110 ---- target/arm/cpu.h | 4 + target/arm/kvm_arm.h | 19 + target/arm/tcg/translate.h | 5 + target/arm/tcg/a64.decode | 152 +++++ hw/arm/sbsa-ref.c | 4 +- hw/arm/vexpress.c | 40 +- hw/arm/virt.c | 73 ++- target/arm/cortex-regs.c | 11 +- target/arm/cpu.c | 9 +- target/arm/debug_helper.c | 11 +- target/arm/kvm.c | 35 + target/arm/kvm64.c | 5 + target/arm/tcg/sve_helper.c | 6 + target/arm/tcg/translate-a64.c | 1321 ++++++++++++++++---------------------- target/arm/tcg/meson.build | 1 + 20 files changed, 979 insertions(+), 924 deletions(-) create mode 100644 docs/system/devices/usb-u2f.rst delete mode 100644 docs/u2f.txt create mode 100644 target/arm/tcg/a64.decode From: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> The world outside moves to newer and newer cpu cores. Let move SBSA Reference Platform to something newer as well. Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com> Message-id: 20230506183417.1360427-1-marcin.juszkiewicz@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/sbsa-ref.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/arm/sbsa-ref.c b/hw/arm/sbsa-ref.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/sbsa-ref.c +++ b/hw/arm/sbsa-ref.c @@ -XXX,XX +XXX,XX @@ static void sbsa_ref_class_init(ObjectClass *oc, void *data) mc->init = sbsa_ref_init; mc->desc = "QEMU 'SBSA Reference' ARM Virtual Machine"; - mc->default_cpu_type = ARM_CPU_TYPE_NAME("cortex-a57"); + mc->default_cpu_type = ARM_CPU_TYPE_NAME("neoverse-n1"); mc->max_cpus = 512; mc->pci_allow_0_address = true; mc->minimum_page_bits = 12; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> If vd == vm, copy vm to scratch, so that we can pre-zero the output and still access the gather indicies. Cc: qemu-stable@nongnu.org Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1612 Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230504104232.1877774-1-richard.henderson@linaro.org Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/sve_helper.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/target/arm/tcg/sve_helper.c b/target/arm/tcg/sve_helper.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/sve_helper.c +++ b/target/arm/tcg/sve_helper.c @@ -XXX,XX +XXX,XX @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm, intptr_t reg_off; SVEHostPage info; target_ulong addr, in_page; + ARMVectorReg scratch; /* Skip to the first true predicate. */ reg_off = find_next_active(vg, 0, reg_max, esz); @@ -XXX,XX +XXX,XX @@ void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm, return; } + /* Protect against overlap between vd and vm. */ + if (unlikely(vd == vm)) { + vm = memcpy(&scratch, vm, reg_max); + } + /* * Probe the first element, allowing faults. */ -- 2.34.1 From: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> At Linaro I work on sbsa-ref, know direction it goes. May not get code details each time. Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-id: 20230515143753.365591-1-marcin.juszkiewicz@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index XXXXXXX..XXXXXXX 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -XXX,XX +XXX,XX @@ SBSA-REF M: Radoslaw Biernacki <rad@semihalf.com> M: Peter Maydell <peter.maydell@linaro.org> R: Leif Lindholm <quic_llindhol@quicinc.com> +R: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> L: qemu-arm@nongnu.org S: Maintained F: hw/arm/sbsa-ref.c -- 2.34.1 From: Cornelia Huck <cohuck@redhat.com> Extend the 'mte' property for the virt machine to cover KVM as well. For KVM, we don't allocate tag memory, but instead enable the capability. If MTE has been enabled, we need to disable migration, as we do not yet have a way to migrate the tags as well. Therefore, MTE will stay off with KVM unless requested explicitly. Signed-off-by: Cornelia Huck <cohuck@redhat.com> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230428095533.21747-2-cohuck@redhat.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/cpu.h | 4 +++ target/arm/kvm_arm.h | 19 ++++++++++++ hw/arm/virt.c | 73 +++++++++++++++++++++++++------------------- target/arm/cpu.c | 9 +++--- target/arm/kvm.c | 35 +++++++++++++++++++++ target/arm/kvm64.c | 5 +++ 6 files changed, 109 insertions(+), 36 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 @@ struct ArchCPU { */ uint32_t psci_conduit; + /* CPU has Memory Tag Extension */ + bool has_mte; + /* For v8M, initial value of the Secure VTOR */ uint32_t init_svtor; /* For v8M, initial value of the Non-secure VTOR */ @@ -XXX,XX +XXX,XX @@ struct ArchCPU { bool prop_pauth; bool prop_pauth_impdef; bool prop_lpa2; + OnOffAuto prop_mte; /* DCZ blocksize, in log_2(words), ie low 4 bits of DCZID_EL0 */ uint32_t dcz_blocksize; diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm_arm.h +++ b/target/arm/kvm_arm.h @@ -XXX,XX +XXX,XX @@ bool kvm_arm_pmu_supported(void); */ bool kvm_arm_sve_supported(void); +/** + * kvm_arm_mte_supported: + * + * Returns: true if KVM can enable MTE, and false otherwise. + */ +bool kvm_arm_mte_supported(void); + /** * kvm_arm_get_max_vm_ipa_size: * @ms: Machine state handle @@ -XXX,XX +XXX,XX @@ void kvm_arm_pvtime_init(CPUState *cs, uint64_t ipa); int kvm_arm_set_irq(int cpu, int irqtype, int irq, int level); +void kvm_arm_enable_mte(Object *cpuobj, Error **errp); + #else /* @@ -XXX,XX +XXX,XX @@ static inline bool kvm_arm_steal_time_supported(void) return false; } +static inline bool kvm_arm_mte_supported(void) +{ + return false; +} + /* * These functions should never actually be called without KVM support. */ @@ -XXX,XX +XXX,XX @@ static inline uint32_t kvm_arm_sve_get_vls(CPUState *cs) g_assert_not_reached(); } +static inline void kvm_arm_enable_mte(Object *cpuobj, Error **errp) +{ + g_assert_not_reached(); +} + #endif static inline const char *gic_class_name(void) diff --git a/hw/arm/virt.c b/hw/arm/virt.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/virt.c +++ b/hw/arm/virt.c @@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine) exit(1); } - if (vms->mte && (kvm_enabled() || hvf_enabled())) { + if (vms->mte && hvf_enabled()) { error_report("mach-virt: %s does not support providing " "MTE to the guest CPU", current_accel_name()); @@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine) } if (vms->mte) { - /* Create the memory region only once, but link to all cpus. */ - if (!tag_sysmem) { - /* - * The property exists only if MemTag is supported. - * If it is, we must allocate the ram to back that up. - */ - if (!object_property_find(cpuobj, "tag-memory")) { - error_report("MTE requested, but not supported " - "by the guest CPU"); + if (tcg_enabled()) { + /* Create the memory region only once, but link to all cpus. */ + if (!tag_sysmem) { + /* + * The property exists only if MemTag is supported. + * If it is, we must allocate the ram to back that up. + */ + if (!object_property_find(cpuobj, "tag-memory")) { + error_report("MTE requested, but not supported " + "by the guest CPU"); + exit(1); + } + + tag_sysmem = g_new(MemoryRegion, 1); + memory_region_init(tag_sysmem, OBJECT(machine), + "tag-memory", UINT64_MAX / 32); + + if (vms->secure) { + secure_tag_sysmem = g_new(MemoryRegion, 1); + memory_region_init(secure_tag_sysmem, OBJECT(machine), + "secure-tag-memory", + UINT64_MAX / 32); + + /* As with ram, secure-tag takes precedence over tag. */ + memory_region_add_subregion_overlap(secure_tag_sysmem, + 0, tag_sysmem, -1); + } + } + + object_property_set_link(cpuobj, "tag-memory", + OBJECT(tag_sysmem), &error_abort); + if (vms->secure) { + object_property_set_link(cpuobj, "secure-tag-memory", + OBJECT(secure_tag_sysmem), + &error_abort); + } + } else if (kvm_enabled()) { + if (!kvm_arm_mte_supported()) { + error_report("MTE requested, but not supported by KVM"); exit(1); } - - tag_sysmem = g_new(MemoryRegion, 1); - memory_region_init(tag_sysmem, OBJECT(machine), - "tag-memory", UINT64_MAX / 32); - - if (vms->secure) { - secure_tag_sysmem = g_new(MemoryRegion, 1); - memory_region_init(secure_tag_sysmem, OBJECT(machine), - "secure-tag-memory", UINT64_MAX / 32); - - /* As with ram, secure-tag takes precedence over tag. */ - memory_region_add_subregion_overlap(secure_tag_sysmem, 0, - tag_sysmem, -1); - } - } - - object_property_set_link(cpuobj, "tag-memory", OBJECT(tag_sysmem), - &error_abort); - if (vms->secure) { - object_property_set_link(cpuobj, "secure-tag-memory", - OBJECT(secure_tag_sysmem), - &error_abort); + kvm_arm_enable_mte(cpuobj, &error_abort); } } 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 @@ void arm_cpu_post_init(Object *obj) qdev_prop_allow_set_link_before_realize, OBJ_PROP_LINK_STRONG); } + cpu->has_mte = true; } #endif } @@ -XXX,XX +XXX,XX @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp) } if (cpu->tag_memory) { error_setg(errp, - "Cannot enable %s when guest CPUs has MTE enabled", + "Cannot enable %s when guest CPUs has tag memory enabled", current_accel_name()); return; } @@ -XXX,XX +XXX,XX @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp) } #ifndef CONFIG_USER_ONLY - if (cpu->tag_memory == NULL && cpu_isar_feature(aa64_mte, cpu)) { + if (!cpu->has_mte && cpu_isar_feature(aa64_mte, cpu)) { /* - * Disable the MTE feature bits if we do not have tag-memory - * provided by the machine. + * Disable the MTE feature bits if we do not have the feature + * setup by the machine. */ cpu->isar.id_aa64pfr1 = FIELD_DP64(cpu->isar.id_aa64pfr1, ID_AA64PFR1, MTE, 0); diff --git a/target/arm/kvm.c b/target/arm/kvm.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm.c +++ b/target/arm/kvm.c @@ -XXX,XX +XXX,XX @@ #include "hw/boards.h" #include "hw/irq.h" #include "qemu/log.h" +#include "migration/blocker.h" const KVMCapabilityInfo kvm_arch_required_capabilities[] = { KVM_CAP_LAST_INFO @@ -XXX,XX +XXX,XX @@ bool kvm_arch_cpu_check_are_resettable(void) void kvm_arch_accel_class_init(ObjectClass *oc) { } + +void kvm_arm_enable_mte(Object *cpuobj, Error **errp) +{ + static bool tried_to_enable; + static bool succeeded_to_enable; + Error *mte_migration_blocker = NULL; + int ret; + + if (!tried_to_enable) { + /* + * MTE on KVM is enabled on a per-VM basis (and retrying doesn't make + * sense), and we only want a single migration blocker as well. + */ + tried_to_enable = true; + + ret = kvm_vm_enable_cap(kvm_state, KVM_CAP_ARM_MTE, 0); + if (ret) { + error_setg_errno(errp, -ret, "Failed to enable KVM_CAP_ARM_MTE"); + return; + } + + /* TODO: add proper migration support with MTE enabled */ + error_setg(&mte_migration_blocker, + "Live migration disabled due to MTE enabled"); + if (migrate_add_blocker(mte_migration_blocker, errp)) { + error_free(mte_migration_blocker); + return; + } + succeeded_to_enable = true; + } + if (succeeded_to_enable) { + object_property_set_bool(cpuobj, "has_mte", true, NULL); + } +} diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/kvm64.c +++ b/target/arm/kvm64.c @@ -XXX,XX +XXX,XX @@ bool kvm_arm_steal_time_supported(void) return kvm_check_extension(kvm_state, KVM_CAP_STEAL_TIME); } +bool kvm_arm_mte_supported(void) +{ + return kvm_check_extension(kvm_state, KVM_CAP_ARM_MTE); +} + QEMU_BUILD_BUG_ON(KVM_ARM64_SVE_VQ_MIN != 1); uint32_t kvm_arm_sve_get_vls(CPUState *cs) -- 2.34.1 From: Alex Bennée <alex.bennee@linaro.org> The commit b3aa2f2128 (target/arm: provide stubs for more external debug registers) was added to handle HyperV's unconditional usage of Debug Communications Channel. It turns out that Linux will similarly break if you enable CONFIG_HVC_DCC "ARM JTAG DCC console". Extend the registers we RAZ/WI set to avoid this. Cc: Anders Roxell <anders.roxell@linaro.org> Cc: Evgeny Iakovlev <eiakovlev@linux.microsoft.com> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230516104420.407912-1-alex.bennee@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/debug_helper.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) 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 const ARMCPRegInfo debug_cp_reginfo[] = { .access = PL0_R, .accessfn = access_tdcc, .type = ARM_CP_CONST, .resetvalue = 0 }, /* - * OSDTRRX_EL1/OSDTRTX_EL1 are used for save and restore of DBGDTRRX_EL0. - * It is a component of the Debug Communications Channel, which is not implemented. + * These registers belong to the Debug Communications Channel, + * which is not implemented. However we implement RAZ/WI behaviour + * with trapping to prevent spurious SIGILLs if the guest OS does + * access them as the support cannot be probed for. */ { .name = "OSDTRRX_EL1", .state = ARM_CP_STATE_BOTH, .cp = 14, .opc0 = 2, .opc1 = 0, .crn = 0, .crm = 0, .opc2 = 2, @@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo debug_cp_reginfo[] = { .opc0 = 2, .opc1 = 0, .crn = 0, .crm = 3, .opc2 = 2, .access = PL1_RW, .accessfn = access_tdcc, .type = ARM_CP_CONST, .resetvalue = 0 }, + /* DBGDTRTX_EL0/DBGDTRRX_EL0 depend on direction */ + { .name = "DBGDTR_EL0", .state = ARM_CP_STATE_BOTH, .cp = 14, + .opc0 = 2, .opc1 = 3, .crn = 0, .crm = 5, .opc2 = 0, + .access = PL0_RW, .accessfn = access_tdcc, + .type = ARM_CP_CONST, .resetvalue = 0 }, /* * OSECCR_EL1 provides a mechanism for an operating system * to access the contents of EDECCR. EDECCR is not implemented though, -- 2.34.1 From: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> Bochs card is normal PCI Express card so it fits better in system with PCI Express bus. VGA is simple legacy PCI card. Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org> Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com> Message-id: 20230505120936.1097060-1-marcin.juszkiewicz@linaro.org Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- hw/arm/sbsa-ref.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/hw/arm/sbsa-ref.c b/hw/arm/sbsa-ref.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/sbsa-ref.c +++ b/hw/arm/sbsa-ref.c @@ -XXX,XX +XXX,XX @@ static void create_pcie(SBSAMachineState *sms) } } - pci_create_simple(pci->bus, -1, "VGA"); + pci_create_simple(pci->bus, -1, "bochs-display"); create_smmu(sms, pci->bus); } -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Split out all of the decode stuff from aarch64_tr_translate_insn. Call it disas_a64_legacy to indicate it will be replaced. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-2-peter.maydell@linaro.org [PMM: Rebased] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/translate-a64.c | 82 ++++++++++++++++++---------------- 1 file changed, 44 insertions(+), 38 deletions(-) diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool btype_destination_ok(uint32_t insn, bool bt, int btype) return false; } +/* C3.1 A64 instruction index by encoding */ +static void disas_a64_legacy(DisasContext *s, uint32_t insn) +{ + switch (extract32(insn, 25, 4)) { + 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: + if (!disas_sve(s, insn)) { + unallocated_encoding(s); + } + break; + case 0x8: case 0x9: /* Data processing - immediate */ + disas_data_proc_imm(s, insn); + break; + case 0xa: case 0xb: /* Branch, exception generation and system insns */ + disas_b_exc_sys(s, insn); + break; + case 0x4: + case 0x6: + case 0xc: + case 0xe: /* Loads and stores */ + disas_ldst(s, insn); + break; + case 0x5: + case 0xd: /* Data processing - register */ + disas_data_proc_reg(s, insn); + break; + case 0x7: + case 0xf: /* Data processing - SIMD and floating point */ + disas_data_proc_simd_fp(s, insn); + break; + default: + assert(FALSE); /* all 15 cases should be handled above */ + break; + } +} + static void aarch64_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cpu) { @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) disas_sme_fa64(s, insn); } - switch (extract32(insn, 25, 4)) { - 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: - if (!disas_sve(s, insn)) { - unallocated_encoding(s); - } - break; - case 0x8: case 0x9: /* Data processing - immediate */ - disas_data_proc_imm(s, insn); - break; - case 0xa: case 0xb: /* Branch, exception generation and system insns */ - disas_b_exc_sys(s, insn); - break; - case 0x4: - case 0x6: - case 0xc: - case 0xe: /* Loads and stores */ - disas_ldst(s, insn); - break; - case 0x5: - case 0xd: /* Data processing - register */ - disas_data_proc_reg(s, insn); - break; - case 0x7: - case 0xf: /* Data processing - SIMD and floating point */ - disas_data_proc_simd_fp(s, insn); - break; - default: - assert(FALSE); /* all 15 cases should be handled above */ - break; - } + disas_a64_legacy(s, insn); /* * After execution of most insns, btype is reset to 0. -- 2.34.1 The A64 translator uses a hand-written decoder for everything except SVE or SME. It's fairly well structured, but it's becoming obvious that it's still more painful to add instructions to than the A32 translator, because putting a new instruction into the right place in a hand-written decoder is much harder than adding new instruction patterns to a decodetree file. As the first step in conversion to decodetree, create the skeleton of the decodetree decoder; where it does not handle instructions we will fall back to the legacy decoder (which will be for everything at the moment, since there are no patterns in a64.decode). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-3-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 20 ++++++++++++++++++++ target/arm/tcg/translate-a64.c | 18 +++++++++++------- target/arm/tcg/meson.build | 1 + 3 files changed, 32 insertions(+), 7 deletions(-) create mode 100644 target/arm/tcg/a64.decode diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ +# AArch64 A64 allowed instruction decoding +# +# Copyright (c) 2023 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/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ enum a64_shift_type { A64_SHIFT_TYPE_ROR = 3 }; +/* + * Include the generated decoders. + */ + +#include "decode-sme-fa64.c.inc" +#include "decode-a64.c.inc" + /* Table based decoder typedefs - used when the relevant bits for decode * are too awkwardly scattered across the instruction (eg SIMD). */ @@ -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; @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) disas_sme_fa64(s, insn); } - disas_a64_legacy(s, insn); + + if (!disas_a64(s, insn)) { + disas_a64_legacy(s, insn); + } /* * After execution of most insns, btype is reset to 0. diff --git a/target/arm/tcg/meson.build b/target/arm/tcg/meson.build index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/meson.build +++ b/target/arm/tcg/meson.build @@ -XXX,XX +XXX,XX @@ gen = [ decodetree.process('a32-uncond.decode', extra_args: '--static-decode=disas_a32_uncond'), decodetree.process('t32.decode', extra_args: '--static-decode=disas_t32'), decodetree.process('t16.decode', extra_args: ['-w', '16', '--static-decode=disas_t16']), + decodetree.process('a64.decode', extra_args: ['--static-decode=disas_a64']), ] arm_ss.add(gen) -- 2.34.1 The SVE and SME decode is already done by decodetree. Pull the calls to these decoders out of the legacy decoder. This doesn't change behaviour because all the patterns in sve.decode and sme.decode already require the bits that the legacy decoder is decoding to have the correct values. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-4-peter.maydell@linaro.org --- target/arm/tcg/translate-a64.c | 20 ++++---------------- 1 file changed, 4 insertions(+), 16 deletions(-) diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool btype_destination_ok(uint32_t insn, bool bt, int btype) static void disas_a64_legacy(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 4)) { - 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: - if (!disas_sve(s, insn)) { - unallocated_encoding(s); - } - break; case 0x8: case 0x9: /* Data processing - immediate */ disas_data_proc_imm(s, insn); break; @@ -XXX,XX +XXX,XX @@ static void disas_a64_legacy(DisasContext *s, uint32_t insn) disas_data_proc_simd_fp(s, insn); break; default: - assert(FALSE); /* all 15 cases should be handled above */ + unallocated_encoding(s); break; } } @@ -XXX,XX +XXX,XX @@ static void aarch64_tr_translate_insn(DisasContextBase *dcbase, CPUState *cpu) disas_sme_fa64(s, insn); } - - if (!disas_a64(s, insn)) { + if (!disas_a64(s, insn) && + !disas_sme(s, insn) && + !disas_sve(s, insn)) { disas_a64_legacy(s, insn); } -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the ADR and ADRP instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-5-peter.maydell@linaro.org [PMM: Rebased] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/a64.decode | 13 ++++++++++++ target/arm/tcg/translate-a64.c | 38 +++++++++++++--------------------- 2 files changed, 27 insertions(+), 24 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ # # This file is processed by scripts/decodetree.py # + +&ri rd imm + + +### Data Processing - Immediate + +# PC-rel addressing + +%imm_pcrel 5:s19 29:2 +@pcrel . .. ..... ................... rd:5 &ri imm=%imm_pcrel + +ADR 0 .. 10000 ................... ..... @pcrel +ADRP 1 .. 10000 ................... ..... @pcrel diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_ldst(DisasContext *s, uint32_t insn) } } -/* PC-rel. addressing - * 31 30 29 28 24 23 5 4 0 - * +----+-------+-----------+-------------------+------+ - * | op | immlo | 1 0 0 0 0 | immhi | Rd | - * +----+-------+-----------+-------------------+------+ +/* + * PC-rel. addressing */ -static void disas_pc_rel_adr(DisasContext *s, uint32_t insn) + +static bool trans_ADR(DisasContext *s, arg_ri *a) { - unsigned int page, rd; - int64_t offset; + gen_pc_plus_diff(s, cpu_reg(s, a->rd), a->imm); + return true; +} - page = extract32(insn, 31, 1); - /* SignExtend(immhi:immlo) -> offset */ - offset = sextract64(insn, 5, 19); - offset = offset << 2 | extract32(insn, 29, 2); - rd = extract32(insn, 0, 5); +static bool trans_ADRP(DisasContext *s, arg_ri *a) +{ + int64_t offset = (int64_t)a->imm << 12; - if (page) { - /* ADRP (page based) */ - offset <<= 12; - /* The page offset is ok for CF_PCREL. */ - offset -= s->pc_curr & 0xfff; - } - - gen_pc_plus_diff(s, cpu_reg(s, rd), offset); + /* The page offset is ok for CF_PCREL. */ + offset -= s->pc_curr & 0xfff; + gen_pc_plus_diff(s, cpu_reg(s, a->rd), offset); + return true; } /* @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x20: case 0x21: /* PC-rel. addressing */ - disas_pc_rel_adr(s, insn); - break; case 0x22: /* Add/subtract (immediate) */ disas_add_sub_imm(s, insn); break; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Split out specific 32-bit and 64-bit functions. These carry the same signature as tcg_gen_add_i64, and so will be easier to pass as callbacks. Retain gen_add_CC and gen_sub_CC during conversion. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-6-peter.maydell@linaro.org [PMM: rebased] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/translate-a64.c | 149 +++++++++++++++++++-------------- 1 file changed, 84 insertions(+), 65 deletions(-) diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static inline void gen_logic_CC(int sf, TCGv_i64 result) } /* dest = T0 + T1; compute C, N, V and Z flags */ +static void gen_add64_CC(TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + TCGv_i64 result, flag, tmp; + result = tcg_temp_new_i64(); + flag = tcg_temp_new_i64(); + tmp = tcg_temp_new_i64(); + + tcg_gen_movi_i64(tmp, 0); + tcg_gen_add2_i64(result, flag, t0, tmp, t1, tmp); + + tcg_gen_extrl_i64_i32(cpu_CF, flag); + + gen_set_NZ64(result); + + tcg_gen_xor_i64(flag, result, t0); + tcg_gen_xor_i64(tmp, t0, t1); + tcg_gen_andc_i64(flag, flag, tmp); + tcg_gen_extrh_i64_i32(cpu_VF, flag); + + tcg_gen_mov_i64(dest, result); +} + +static void gen_add32_CC(TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + TCGv_i32 t0_32 = tcg_temp_new_i32(); + TCGv_i32 t1_32 = tcg_temp_new_i32(); + TCGv_i32 tmp = tcg_temp_new_i32(); + + tcg_gen_movi_i32(tmp, 0); + tcg_gen_extrl_i64_i32(t0_32, t0); + tcg_gen_extrl_i64_i32(t1_32, t1); + tcg_gen_add2_i32(cpu_NF, cpu_CF, t0_32, tmp, t1_32, tmp); + tcg_gen_mov_i32(cpu_ZF, cpu_NF); + tcg_gen_xor_i32(cpu_VF, cpu_NF, t0_32); + tcg_gen_xor_i32(tmp, t0_32, t1_32); + tcg_gen_andc_i32(cpu_VF, cpu_VF, tmp); + tcg_gen_extu_i32_i64(dest, cpu_NF); +} + static void gen_add_CC(int sf, TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) { if (sf) { - TCGv_i64 result, flag, tmp; - result = tcg_temp_new_i64(); - flag = tcg_temp_new_i64(); - tmp = tcg_temp_new_i64(); - - tcg_gen_movi_i64(tmp, 0); - tcg_gen_add2_i64(result, flag, t0, tmp, t1, tmp); - - tcg_gen_extrl_i64_i32(cpu_CF, flag); - - gen_set_NZ64(result); - - tcg_gen_xor_i64(flag, result, t0); - tcg_gen_xor_i64(tmp, t0, t1); - tcg_gen_andc_i64(flag, flag, tmp); - tcg_gen_extrh_i64_i32(cpu_VF, flag); - - tcg_gen_mov_i64(dest, result); + gen_add64_CC(dest, t0, t1); } else { - /* 32 bit arithmetic */ - TCGv_i32 t0_32 = tcg_temp_new_i32(); - TCGv_i32 t1_32 = tcg_temp_new_i32(); - TCGv_i32 tmp = tcg_temp_new_i32(); - - tcg_gen_movi_i32(tmp, 0); - tcg_gen_extrl_i64_i32(t0_32, t0); - tcg_gen_extrl_i64_i32(t1_32, t1); - tcg_gen_add2_i32(cpu_NF, cpu_CF, t0_32, tmp, t1_32, tmp); - tcg_gen_mov_i32(cpu_ZF, cpu_NF); - tcg_gen_xor_i32(cpu_VF, cpu_NF, t0_32); - tcg_gen_xor_i32(tmp, t0_32, t1_32); - tcg_gen_andc_i32(cpu_VF, cpu_VF, tmp); - tcg_gen_extu_i32_i64(dest, cpu_NF); + gen_add32_CC(dest, t0, t1); } } /* dest = T0 - T1; compute C, N, V and Z flags */ +static void gen_sub64_CC(TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + /* 64 bit arithmetic */ + TCGv_i64 result, flag, tmp; + + result = tcg_temp_new_i64(); + flag = tcg_temp_new_i64(); + tcg_gen_sub_i64(result, t0, t1); + + gen_set_NZ64(result); + + tcg_gen_setcond_i64(TCG_COND_GEU, flag, t0, t1); + tcg_gen_extrl_i64_i32(cpu_CF, flag); + + tcg_gen_xor_i64(flag, result, t0); + tmp = tcg_temp_new_i64(); + tcg_gen_xor_i64(tmp, t0, t1); + tcg_gen_and_i64(flag, flag, tmp); + tcg_gen_extrh_i64_i32(cpu_VF, flag); + tcg_gen_mov_i64(dest, result); +} + +static void gen_sub32_CC(TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) +{ + /* 32 bit arithmetic */ + TCGv_i32 t0_32 = tcg_temp_new_i32(); + TCGv_i32 t1_32 = tcg_temp_new_i32(); + TCGv_i32 tmp; + + tcg_gen_extrl_i64_i32(t0_32, t0); + tcg_gen_extrl_i64_i32(t1_32, t1); + tcg_gen_sub_i32(cpu_NF, t0_32, t1_32); + tcg_gen_mov_i32(cpu_ZF, cpu_NF); + tcg_gen_setcond_i32(TCG_COND_GEU, cpu_CF, t0_32, t1_32); + tcg_gen_xor_i32(cpu_VF, cpu_NF, t0_32); + tmp = tcg_temp_new_i32(); + tcg_gen_xor_i32(tmp, t0_32, t1_32); + tcg_gen_and_i32(cpu_VF, cpu_VF, tmp); + tcg_gen_extu_i32_i64(dest, cpu_NF); +} + static void gen_sub_CC(int sf, TCGv_i64 dest, TCGv_i64 t0, TCGv_i64 t1) { if (sf) { - /* 64 bit arithmetic */ - TCGv_i64 result, flag, tmp; - - result = tcg_temp_new_i64(); - flag = tcg_temp_new_i64(); - tcg_gen_sub_i64(result, t0, t1); - - gen_set_NZ64(result); - - tcg_gen_setcond_i64(TCG_COND_GEU, flag, t0, t1); - tcg_gen_extrl_i64_i32(cpu_CF, flag); - - tcg_gen_xor_i64(flag, result, t0); - tmp = tcg_temp_new_i64(); - tcg_gen_xor_i64(tmp, t0, t1); - tcg_gen_and_i64(flag, flag, tmp); - tcg_gen_extrh_i64_i32(cpu_VF, flag); - tcg_gen_mov_i64(dest, result); + gen_sub64_CC(dest, t0, t1); } else { - /* 32 bit arithmetic */ - TCGv_i32 t0_32 = tcg_temp_new_i32(); - TCGv_i32 t1_32 = tcg_temp_new_i32(); - TCGv_i32 tmp; - - tcg_gen_extrl_i64_i32(t0_32, t0); - tcg_gen_extrl_i64_i32(t1_32, t1); - tcg_gen_sub_i32(cpu_NF, t0_32, t1_32); - tcg_gen_mov_i32(cpu_ZF, cpu_NF); - tcg_gen_setcond_i32(TCG_COND_GEU, cpu_CF, t0_32, t1_32); - tcg_gen_xor_i32(cpu_VF, cpu_NF, t0_32); - tmp = tcg_temp_new_i32(); - tcg_gen_xor_i32(tmp, t0_32, t1_32); - tcg_gen_and_i32(cpu_VF, cpu_VF, tmp); - tcg_gen_extu_i32_i64(dest, cpu_NF); + gen_sub32_CC(dest, t0, t1); } } -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the ADD and SUB (immediate) instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-7-peter.maydell@linaro.org [PMM: Rebased; adjusted to use translate.h's TRANS macro] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/translate.h | 5 +++ target/arm/tcg/a64.decode | 17 ++++++++ target/arm/tcg/translate-a64.c | 73 ++++++++++------------------------ 3 files changed, 42 insertions(+), 53 deletions(-) diff --git a/target/arm/tcg/translate.h b/target/arm/tcg/translate.h index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate.h +++ b/target/arm/tcg/translate.h @@ -XXX,XX +XXX,XX @@ static inline int rsub_8(DisasContext *s, int x) return 8 - x; } +static inline int shl_12(DisasContext *s, int x) +{ + return x << 12; +} + static inline int neon_3same_fp_size(DisasContext *s, int x) { /* Convert 0==fp32, 1==fp16 into a MO_* value */ diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ # &ri rd imm +&rri_sf rd rn imm sf ### Data Processing - Immediate @@ -XXX,XX +XXX,XX @@ ADR 0 .. 10000 ................... ..... @pcrel ADRP 1 .. 10000 ................... ..... @pcrel + +# Add/subtract (immediate) + +%imm12_sh12 10:12 !function=shl_12 +@addsub_imm sf:1 .. ...... . imm:12 rn:5 rd:5 +@addsub_imm12 sf:1 .. ...... . ............ rn:5 rd:5 imm=%imm12_sh12 + +ADD_i . 00 100010 0 ............ ..... ..... @addsub_imm +ADD_i . 00 100010 1 ............ ..... ..... @addsub_imm12 +ADDS_i . 01 100010 0 ............ ..... ..... @addsub_imm +ADDS_i . 01 100010 1 ............ ..... ..... @addsub_imm12 + +SUB_i . 10 100010 0 ............ ..... ..... @addsub_imm +SUB_i . 10 100010 1 ............ ..... ..... @addsub_imm12 +SUBS_i . 11 100010 0 ............ ..... ..... @addsub_imm +SUBS_i . 11 100010 1 ............ ..... ..... @addsub_imm12 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static void disas_ldst(DisasContext *s, uint32_t insn) } } +typedef void ArithTwoOp(TCGv_i64, TCGv_i64, TCGv_i64); + +static bool gen_rri(DisasContext *s, arg_rri_sf *a, + bool rd_sp, bool rn_sp, ArithTwoOp *fn) +{ + TCGv_i64 tcg_rn = rn_sp ? cpu_reg_sp(s, a->rn) : cpu_reg(s, a->rn); + TCGv_i64 tcg_rd = rd_sp ? cpu_reg_sp(s, a->rd) : cpu_reg(s, a->rd); + TCGv_i64 tcg_imm = tcg_constant_i64(a->imm); + + fn(tcg_rd, tcg_rn, tcg_imm); + if (!a->sf) { + tcg_gen_ext32u_i64(tcg_rd, tcg_rd); + } + return true; +} + /* * PC-rel. addressing */ @@ -XXX,XX +XXX,XX @@ static bool trans_ADRP(DisasContext *s, arg_ri *a) /* * Add/subtract (immediate) - * - * 31 30 29 28 23 22 21 10 9 5 4 0 - * +--+--+--+-------------+--+-------------+-----+-----+ - * |sf|op| S| 1 0 0 0 1 0 |sh| imm12 | Rn | Rd | - * +--+--+--+-------------+--+-------------+-----+-----+ - * - * sf: 0 -> 32bit, 1 -> 64bit - * op: 0 -> add , 1 -> sub - * S: 1 -> set flags - * sh: 1 -> LSL imm by 12 */ -static void disas_add_sub_imm(DisasContext *s, uint32_t insn) -{ - int rd = extract32(insn, 0, 5); - int rn = extract32(insn, 5, 5); - uint64_t imm = extract32(insn, 10, 12); - bool shift = extract32(insn, 22, 1); - bool setflags = extract32(insn, 29, 1); - bool sub_op = extract32(insn, 30, 1); - bool is_64bit = extract32(insn, 31, 1); - - TCGv_i64 tcg_rn = cpu_reg_sp(s, rn); - TCGv_i64 tcg_rd = setflags ? cpu_reg(s, rd) : cpu_reg_sp(s, rd); - TCGv_i64 tcg_result; - - if (shift) { - imm <<= 12; - } - - tcg_result = tcg_temp_new_i64(); - if (!setflags) { - if (sub_op) { - tcg_gen_subi_i64(tcg_result, tcg_rn, imm); - } else { - tcg_gen_addi_i64(tcg_result, tcg_rn, imm); - } - } else { - TCGv_i64 tcg_imm = tcg_constant_i64(imm); - if (sub_op) { - gen_sub_CC(is_64bit, tcg_result, tcg_rn, tcg_imm); - } else { - gen_add_CC(is_64bit, tcg_result, tcg_rn, tcg_imm); - } - } - - if (is_64bit) { - tcg_gen_mov_i64(tcg_rd, tcg_result); - } else { - tcg_gen_ext32u_i64(tcg_rd, tcg_result); - } -} +TRANS(ADD_i, gen_rri, a, 1, 1, tcg_gen_add_i64) +TRANS(SUB_i, gen_rri, a, 1, 1, tcg_gen_sub_i64) +TRANS(ADDS_i, gen_rri, a, 0, 1, a->sf ? gen_add64_CC : gen_add32_CC) +TRANS(SUBS_i, gen_rri, a, 0, 1, a->sf ? gen_sub64_CC : gen_sub32_CC) /* * Add/subtract (immediate, with tags) @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x22: /* Add/subtract (immediate) */ - disas_add_sub_imm(s, insn); - break; case 0x23: /* Add/subtract (immediate, with tags) */ disas_add_sub_imm_with_tags(s, insn); break; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the ADDG and SUBG (immediate) instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-8-peter.maydell@linaro.org [PMM: Rebased; use TRANS_FEAT()] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/a64.decode | 8 +++++++ target/arm/tcg/translate-a64.c | 38 ++++++++++------------------------ 2 files changed, 19 insertions(+), 27 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ SUB_i . 10 100010 0 ............ ..... ..... @addsub_imm SUB_i . 10 100010 1 ............ ..... ..... @addsub_imm12 SUBS_i . 11 100010 0 ............ ..... ..... @addsub_imm SUBS_i . 11 100010 1 ............ ..... ..... @addsub_imm12 + +# Add/subtract (immediate with tags) + +&rri_tag rd rn uimm6 uimm4 +@addsub_imm_tag . .. ...... . uimm6:6 .. uimm4:4 rn:5 rd:5 &rri_tag + +ADDG_i 1 00 100011 0 ...... 00 .... ..... ..... @addsub_imm_tag +SUBG_i 1 10 100011 0 ...... 00 .... ..... ..... @addsub_imm_tag diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ TRANS(SUBS_i, gen_rri, a, 0, 1, a->sf ? gen_sub64_CC : gen_sub32_CC) /* * Add/subtract (immediate, with tags) - * - * 31 30 29 28 23 22 21 16 14 10 9 5 4 0 - * +--+--+--+-------------+--+---------+--+-------+-----+-----+ - * |sf|op| S| 1 0 0 0 1 1 |o2| uimm6 |o3| uimm4 | Rn | Rd | - * +--+--+--+-------------+--+---------+--+-------+-----+-----+ - * - * op: 0 -> add, 1 -> sub */ -static void disas_add_sub_imm_with_tags(DisasContext *s, uint32_t insn) + +static bool gen_add_sub_imm_with_tags(DisasContext *s, arg_rri_tag *a, + bool sub_op) { - int rd = extract32(insn, 0, 5); - int rn = extract32(insn, 5, 5); - int uimm4 = extract32(insn, 10, 4); - int uimm6 = extract32(insn, 16, 6); - bool sub_op = extract32(insn, 30, 1); TCGv_i64 tcg_rn, tcg_rd; int imm; - /* Test all of sf=1, S=0, o2=0, o3=0. */ - if ((insn & 0xa040c000u) != 0x80000000u || - !dc_isar_feature(aa64_mte_insn_reg, s)) { - unallocated_encoding(s); - return; - } - - imm = uimm6 << LOG2_TAG_GRANULE; + imm = a->uimm6 << LOG2_TAG_GRANULE; if (sub_op) { imm = -imm; } - tcg_rn = cpu_reg_sp(s, rn); - tcg_rd = cpu_reg_sp(s, rd); + tcg_rn = cpu_reg_sp(s, a->rn); + tcg_rd = cpu_reg_sp(s, a->rd); if (s->ata) { gen_helper_addsubg(tcg_rd, cpu_env, tcg_rn, tcg_constant_i32(imm), - tcg_constant_i32(uimm4)); + tcg_constant_i32(a->uimm4)); } else { tcg_gen_addi_i64(tcg_rd, tcg_rn, imm); gen_address_with_allocation_tag0(tcg_rd, tcg_rd); } + return true; } +TRANS_FEAT(ADDG_i, aa64_mte_insn_reg, gen_add_sub_imm_with_tags, a, false) +TRANS_FEAT(SUBG_i, aa64_mte_insn_reg, gen_add_sub_imm_with_tags, a, true) + /* The input should be a value in the bottom e bits (with higher * bits zero); returns that value replicated into every element * of size e in a 64 bit integer. @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x23: /* Add/subtract (immediate, with tags) */ - disas_add_sub_imm_with_tags(s, insn); - break; case 0x24: /* Logical (immediate) */ disas_logic_imm(s, insn); break; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Use the bitops.h macro rather than rolling our own here. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-9-peter.maydell@linaro.org --- target/arm/tcg/translate-a64.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static uint64_t bitfield_replicate(uint64_t mask, unsigned int e) return mask; } -/* Return a value with the bottom len bits set (where 0 < len <= 64) */ -static inline uint64_t bitmask64(unsigned int length) -{ - assert(length > 0 && length <= 64); - return ~0ULL >> (64 - length); -} - /* Simplified variant of pseudocode DecodeBitMasks() for the case where we * only require the wmask. Returns false if the imms/immr/immn are a reserved * value (ie should cause a guest UNDEF exception), and true if they are @@ -XXX,XX +XXX,XX @@ bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn, /* Create the value of one element: s+1 set bits rotated * by r within the element (which is e bits wide)... */ - mask = bitmask64(s + 1); + mask = MAKE_64BIT_MASK(0, s + 1); if (r) { mask = (mask >> r) | (mask << (e - r)); - mask &= bitmask64(e); + mask &= MAKE_64BIT_MASK(0, e); } /* ...then replicate the element over the whole 64 bit value */ mask = bitfield_replicate(mask, e); -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the ADD, ORR, EOR, ANDS (immediate) instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-10-peter.maydell@linaro.org [PMM: rebased] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/a64.decode | 15 ++++++ target/arm/tcg/translate-a64.c | 94 +++++++++++----------------------- 2 files changed, 44 insertions(+), 65 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ SUBS_i . 11 100010 1 ............ ..... ..... @addsub_imm12 ADDG_i 1 00 100011 0 ...... 00 .... ..... ..... @addsub_imm_tag SUBG_i 1 10 100011 0 ...... 00 .... ..... ..... @addsub_imm_tag + +# Logical (immediate) + +&rri_log rd rn sf dbm +@logic_imm_64 1 .. ...... dbm:13 rn:5 rd:5 &rri_log sf=1 +@logic_imm_32 0 .. ...... 0 dbm:12 rn:5 rd:5 &rri_log sf=0 + +AND_i . 00 100100 . ...... ...... ..... ..... @logic_imm_64 +AND_i . 00 100100 . ...... ...... ..... ..... @logic_imm_32 +ORR_i . 01 100100 . ...... ...... ..... ..... @logic_imm_64 +ORR_i . 01 100100 . ...... ...... ..... ..... @logic_imm_32 +EOR_i . 10 100100 . ...... ...... ..... ..... @logic_imm_64 +EOR_i . 10 100100 . ...... ...... ..... ..... @logic_imm_32 +ANDS_i . 11 100100 . ...... ...... ..... ..... @logic_imm_64 +ANDS_i . 11 100100 . ...... ...... ..... ..... @logic_imm_32 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static uint64_t bitfield_replicate(uint64_t mask, unsigned int e) return mask; } -/* Simplified variant of pseudocode DecodeBitMasks() for the case where we +/* + * Logical (immediate) + */ + +/* + * Simplified variant of pseudocode DecodeBitMasks() for the case where we * only require the wmask. Returns false if the imms/immr/immn are a reserved * value (ie should cause a guest UNDEF exception), and true if they are * valid, in which case the decoded bit pattern is written to result. @@ -XXX,XX +XXX,XX @@ bool logic_imm_decode_wmask(uint64_t *result, unsigned int immn, return true; } -/* Logical (immediate) - * 31 30 29 28 23 22 21 16 15 10 9 5 4 0 - * +----+-----+-------------+---+------+------+------+------+ - * | sf | opc | 1 0 0 1 0 0 | N | immr | imms | Rn | Rd | - * +----+-----+-------------+---+------+------+------+------+ - */ -static void disas_logic_imm(DisasContext *s, uint32_t insn) +static bool gen_rri_log(DisasContext *s, arg_rri_log *a, bool set_cc, + void (*fn)(TCGv_i64, TCGv_i64, int64_t)) { - unsigned int sf, opc, is_n, immr, imms, rn, rd; TCGv_i64 tcg_rd, tcg_rn; - uint64_t wmask; - bool is_and = false; + uint64_t imm; - sf = extract32(insn, 31, 1); - opc = extract32(insn, 29, 2); - is_n = extract32(insn, 22, 1); - immr = extract32(insn, 16, 6); - imms = extract32(insn, 10, 6); - rn = extract32(insn, 5, 5); - rd = extract32(insn, 0, 5); - - if (!sf && is_n) { - unallocated_encoding(s); - return; + /* Some immediate field values are reserved. */ + if (!logic_imm_decode_wmask(&imm, extract32(a->dbm, 12, 1), + extract32(a->dbm, 0, 6), + extract32(a->dbm, 6, 6))) { + return false; + } + if (!a->sf) { + imm &= 0xffffffffull; } - if (opc == 0x3) { /* ANDS */ - tcg_rd = cpu_reg(s, rd); - } else { - tcg_rd = cpu_reg_sp(s, rd); - } - tcg_rn = cpu_reg(s, rn); + tcg_rd = set_cc ? cpu_reg(s, a->rd) : cpu_reg_sp(s, a->rd); + tcg_rn = cpu_reg(s, a->rn); - if (!logic_imm_decode_wmask(&wmask, is_n, imms, immr)) { - /* some immediate field values are reserved */ - unallocated_encoding(s); - return; + fn(tcg_rd, tcg_rn, imm); + if (set_cc) { + gen_logic_CC(a->sf, tcg_rd); } - - if (!sf) { - wmask &= 0xffffffff; - } - - switch (opc) { - case 0x3: /* ANDS */ - case 0x0: /* AND */ - tcg_gen_andi_i64(tcg_rd, tcg_rn, wmask); - is_and = true; - break; - case 0x1: /* ORR */ - tcg_gen_ori_i64(tcg_rd, tcg_rn, wmask); - break; - case 0x2: /* EOR */ - tcg_gen_xori_i64(tcg_rd, tcg_rn, wmask); - break; - default: - assert(FALSE); /* must handle all above */ - break; - } - - if (!sf && !is_and) { - /* zero extend final result; we know we can skip this for AND - * since the immediate had the high 32 bits clear. - */ + if (!a->sf) { tcg_gen_ext32u_i64(tcg_rd, tcg_rd); } - - if (opc == 3) { /* ANDS */ - gen_logic_CC(sf, tcg_rd); - } + return true; } +TRANS(AND_i, gen_rri_log, a, false, tcg_gen_andi_i64) +TRANS(ORR_i, gen_rri_log, a, false, tcg_gen_ori_i64) +TRANS(EOR_i, gen_rri_log, a, false, tcg_gen_xori_i64) +TRANS(ANDS_i, gen_rri_log, a, true, tcg_gen_andi_i64) + /* * Move wide (immediate) * @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x24: /* Logical (immediate) */ - disas_logic_imm(s, insn); - break; case 0x25: /* Move wide (immediate) */ disas_movw_imm(s, insn); break; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the MON, MOVZ, MOVK instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-11-peter.maydell@linaro.org [PMM: Rebased] Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/a64.decode | 13 ++++++ target/arm/tcg/translate-a64.c | 73 ++++++++++++++-------------------- 2 files changed, 42 insertions(+), 44 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ EOR_i . 10 100100 . ...... ...... ..... ..... @logic_imm_64 EOR_i . 10 100100 . ...... ...... ..... ..... @logic_imm_32 ANDS_i . 11 100100 . ...... ...... ..... ..... @logic_imm_64 ANDS_i . 11 100100 . ...... ...... ..... ..... @logic_imm_32 + +# Move wide (immediate) + +&movw rd sf imm hw +@movw_64 1 .. ...... hw:2 imm:16 rd:5 &movw sf=1 +@movw_32 0 .. ...... 0 hw:1 imm:16 rd:5 &movw sf=0 + +MOVN . 00 100101 .. ................ ..... @movw_64 +MOVN . 00 100101 .. ................ ..... @movw_32 +MOVZ . 10 100101 .. ................ ..... @movw_64 +MOVZ . 10 100101 .. ................ ..... @movw_32 +MOVK . 11 100101 .. ................ ..... @movw_64 +MOVK . 11 100101 .. ................ ..... @movw_32 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ TRANS(ANDS_i, gen_rri_log, a, true, tcg_gen_andi_i64) /* * Move wide (immediate) - * - * 31 30 29 28 23 22 21 20 5 4 0 - * +--+-----+-------------+-----+----------------+------+ - * |sf| opc | 1 0 0 1 0 1 | hw | imm16 | Rd | - * +--+-----+-------------+-----+----------------+------+ - * - * sf: 0 -> 32 bit, 1 -> 64 bit - * opc: 00 -> N, 10 -> Z, 11 -> K - * hw: shift/16 (0,16, and sf only 32, 48) */ -static void disas_movw_imm(DisasContext *s, uint32_t insn) + +static bool trans_MOVZ(DisasContext *s, arg_movw *a) { - int rd = extract32(insn, 0, 5); - uint64_t imm = extract32(insn, 5, 16); - int sf = extract32(insn, 31, 1); - int opc = extract32(insn, 29, 2); - int pos = extract32(insn, 21, 2) << 4; - TCGv_i64 tcg_rd = cpu_reg(s, rd); + int pos = a->hw << 4; + tcg_gen_movi_i64(cpu_reg(s, a->rd), (uint64_t)a->imm << pos); + return true; +} - if (!sf && (pos >= 32)) { - unallocated_encoding(s); - return; - } +static bool trans_MOVN(DisasContext *s, arg_movw *a) +{ + int pos = a->hw << 4; + uint64_t imm = a->imm; - switch (opc) { - case 0: /* MOVN */ - case 2: /* MOVZ */ - imm <<= pos; - if (opc == 0) { - imm = ~imm; - } - if (!sf) { - imm &= 0xffffffffu; - } - tcg_gen_movi_i64(tcg_rd, imm); - break; - case 3: /* MOVK */ - tcg_gen_deposit_i64(tcg_rd, tcg_rd, tcg_constant_i64(imm), pos, 16); - if (!sf) { - tcg_gen_ext32u_i64(tcg_rd, tcg_rd); - } - break; - default: - unallocated_encoding(s); - break; + imm = ~(imm << pos); + if (!a->sf) { + imm = (uint32_t)imm; } + tcg_gen_movi_i64(cpu_reg(s, a->rd), imm); + return true; +} + +static bool trans_MOVK(DisasContext *s, arg_movw *a) +{ + int pos = a->hw << 4; + TCGv_i64 tcg_rd, tcg_im; + + tcg_rd = cpu_reg(s, a->rd); + tcg_im = tcg_constant_i64(a->imm); + tcg_gen_deposit_i64(tcg_rd, tcg_rd, tcg_im, pos, 16); + if (!a->sf) { + tcg_gen_ext32u_i64(tcg_rd, tcg_rd); + } + return true; } /* Bitfield @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x25: /* Move wide (immediate) */ - disas_movw_imm(s, insn); - break; case 0x26: /* Bitfield */ disas_bitfield(s, insn); break; -- 2.34.1 From: Richard Henderson <richard.henderson@linaro.org> Convert the BFM, SBFM, UBFM instructions. Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Message-id: 20230512144106.3608981-12-peter.maydell@linaro.org [PMM: Rebased] Signed-off-by: Peter Maydell <peter.maydell@linaro.org> --- target/arm/tcg/a64.decode | 13 +++ target/arm/tcg/translate-a64.c | 144 ++++++++++++++++++--------------- 2 files changed, 94 insertions(+), 63 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ MOVZ . 10 100101 .. ................ ..... @movw_64 MOVZ . 10 100101 .. ................ ..... @movw_32 MOVK . 11 100101 .. ................ ..... @movw_64 MOVK . 11 100101 .. ................ ..... @movw_32 + +# Bitfield + +&bitfield rd rn sf immr imms +@bitfield_64 1 .. ...... 1 immr:6 imms:6 rn:5 rd:5 &bitfield sf=1 +@bitfield_32 0 .. ...... 0 0 immr:5 0 imms:5 rn:5 rd:5 &bitfield sf=0 + +SBFM . 00 100110 . ...... ...... ..... ..... @bitfield_64 +SBFM . 00 100110 . ...... ...... ..... ..... @bitfield_32 +BFM . 01 100110 . ...... ...... ..... ..... @bitfield_64 +BFM . 01 100110 . ...... ...... ..... ..... @bitfield_32 +UBFM . 10 100110 . ...... ...... ..... ..... @bitfield_64 +UBFM . 10 100110 . ...... ...... ..... ..... @bitfield_32 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_MOVK(DisasContext *s, arg_movw *a) return true; } -/* Bitfield - * 31 30 29 28 23 22 21 16 15 10 9 5 4 0 - * +----+-----+-------------+---+------+------+------+------+ - * | sf | opc | 1 0 0 1 1 0 | N | immr | imms | Rn | Rd | - * +----+-----+-------------+---+------+------+------+------+ +/* + * Bitfield */ -static void disas_bitfield(DisasContext *s, uint32_t insn) + +static bool trans_SBFM(DisasContext *s, arg_SBFM *a) { - unsigned int sf, n, opc, ri, si, rn, rd, bitsize, pos, len; - TCGv_i64 tcg_rd, tcg_tmp; + TCGv_i64 tcg_rd = cpu_reg(s, a->rd); + TCGv_i64 tcg_tmp = read_cpu_reg(s, a->rn, 1); + unsigned int bitsize = a->sf ? 64 : 32; + unsigned int ri = a->immr; + unsigned int si = a->imms; + unsigned int pos, len; - sf = extract32(insn, 31, 1); - opc = extract32(insn, 29, 2); - n = extract32(insn, 22, 1); - ri = extract32(insn, 16, 6); - si = extract32(insn, 10, 6); - rn = extract32(insn, 5, 5); - rd = extract32(insn, 0, 5); - bitsize = sf ? 64 : 32; - - if (sf != n || ri >= bitsize || si >= bitsize || opc > 2) { - unallocated_encoding(s); - return; - } - - tcg_rd = cpu_reg(s, rd); - - /* Suppress the zero-extend for !sf. Since RI and SI are constrained - to be smaller than bitsize, we'll never reference data outside the - low 32-bits anyway. */ - tcg_tmp = read_cpu_reg(s, rn, 1); - - /* Recognize simple(r) extractions. */ if (si >= ri) { /* Wd<s-r:0> = Wn<s:r> */ len = (si - ri) + 1; - if (opc == 0) { /* SBFM: ASR, SBFX, SXTB, SXTH, SXTW */ - tcg_gen_sextract_i64(tcg_rd, tcg_tmp, ri, len); - goto done; - } else if (opc == 2) { /* UBFM: UBFX, LSR, UXTB, UXTH */ - tcg_gen_extract_i64(tcg_rd, tcg_tmp, ri, len); - return; + tcg_gen_sextract_i64(tcg_rd, tcg_tmp, ri, len); + if (!a->sf) { + tcg_gen_ext32u_i64(tcg_rd, tcg_rd); } - /* opc == 1, BFXIL fall through to deposit */ + } else { + /* Wd<32+s-r,32-r> = Wn<s:0> */ + len = si + 1; + pos = (bitsize - ri) & (bitsize - 1); + + if (len < ri) { + /* + * Sign extend the destination field from len to fill the + * balance of the word. Let the deposit below insert all + * of those sign bits. + */ + tcg_gen_sextract_i64(tcg_tmp, tcg_tmp, 0, len); + len = ri; + } + + /* + * We start with zero, and we haven't modified any bits outside + * bitsize, therefore no final zero-extension is unneeded for !sf. + */ + tcg_gen_deposit_z_i64(tcg_rd, tcg_tmp, pos, len); + } + return true; +} + +static bool trans_UBFM(DisasContext *s, arg_UBFM *a) +{ + TCGv_i64 tcg_rd = cpu_reg(s, a->rd); + TCGv_i64 tcg_tmp = read_cpu_reg(s, a->rn, 1); + unsigned int bitsize = a->sf ? 64 : 32; + unsigned int ri = a->immr; + unsigned int si = a->imms; + unsigned int pos, len; + + tcg_rd = cpu_reg(s, a->rd); + tcg_tmp = read_cpu_reg(s, a->rn, 1); + + if (si >= ri) { + /* Wd<s-r:0> = Wn<s:r> */ + len = (si - ri) + 1; + tcg_gen_extract_i64(tcg_rd, tcg_tmp, ri, len); + } else { + /* Wd<32+s-r,32-r> = Wn<s:0> */ + len = si + 1; + pos = (bitsize - ri) & (bitsize - 1); + tcg_gen_deposit_z_i64(tcg_rd, tcg_tmp, pos, len); + } + return true; +} + +static bool trans_BFM(DisasContext *s, arg_BFM *a) +{ + TCGv_i64 tcg_rd = cpu_reg(s, a->rd); + TCGv_i64 tcg_tmp = read_cpu_reg(s, a->rn, 1); + unsigned int bitsize = a->sf ? 64 : 32; + unsigned int ri = a->immr; + unsigned int si = a->imms; + unsigned int pos, len; + + tcg_rd = cpu_reg(s, a->rd); + tcg_tmp = read_cpu_reg(s, a->rn, 1); + + if (si >= ri) { + /* Wd<s-r:0> = Wn<s:r> */ tcg_gen_shri_i64(tcg_tmp, tcg_tmp, ri); + len = (si - ri) + 1; pos = 0; } else { - /* Handle the ri > si case with a deposit - * Wd<32+s-r,32-r> = Wn<s:0> - */ + /* Wd<32+s-r,32-r> = Wn<s:0> */ len = si + 1; pos = (bitsize - ri) & (bitsize - 1); } - if (opc == 0 && len < ri) { - /* SBFM: sign extend the destination field from len to fill - the balance of the word. Let the deposit below insert all - of those sign bits. */ - tcg_gen_sextract_i64(tcg_tmp, tcg_tmp, 0, len); - len = ri; - } - - if (opc == 1) { /* BFM, BFXIL */ - tcg_gen_deposit_i64(tcg_rd, tcg_rd, tcg_tmp, pos, len); - } else { - /* SBFM or UBFM: We start with zero, and we haven't modified - any bits outside bitsize, therefore the zero-extension - below is unneeded. */ - tcg_gen_deposit_z_i64(tcg_rd, tcg_tmp, pos, len); - return; - } - - done: - if (!sf) { /* zero extend final result */ + tcg_gen_deposit_i64(tcg_rd, tcg_rd, tcg_tmp, pos, len); + if (!a->sf) { tcg_gen_ext32u_i64(tcg_rd, tcg_rd); } + return true; } /* Extract @@ -XXX,XX +XXX,XX @@ static void disas_extract(DisasContext *s, uint32_t insn) static void disas_data_proc_imm(DisasContext *s, uint32_t insn) { switch (extract32(insn, 23, 6)) { - case 0x26: /* Bitfield */ - disas_bitfield(s, insn); - break; case 0x27: /* Extract */ disas_extract(s, insn); break; -- 2.34.1 Convert the EXTR instruction to decodetree (this is the only one in the 'Extract" class). This is the last of the dp-immediate insns in the legacy decoder, so we can now remove disas_data_proc_imm(). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-13-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 7 +++ target/arm/tcg/translate-a64.c | 94 +++++++++++----------------------- 2 files changed, 36 insertions(+), 65 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ BFM . 01 100110 . ...... ...... ..... ..... @bitfield_64 BFM . 01 100110 . ...... ...... ..... ..... @bitfield_32 UBFM . 10 100110 . ...... ...... ..... ..... @bitfield_64 UBFM . 10 100110 . ...... ...... ..... ..... @bitfield_32 + +# Extract + +&extract rd rn rm imm sf + +EXTR 1 00 100111 1 0 rm:5 imm:6 rn:5 rd:5 &extract sf=1 +EXTR 0 00 100111 0 0 rm:5 0 imm:5 rn:5 rd:5 &extract sf=0 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_BFM(DisasContext *s, arg_BFM *a) return true; } -/* Extract - * 31 30 29 28 23 22 21 20 16 15 10 9 5 4 0 - * +----+------+-------------+---+----+------+--------+------+------+ - * | sf | op21 | 1 0 0 1 1 1 | N | o0 | Rm | imms | Rn | Rd | - * +----+------+-------------+---+----+------+--------+------+------+ - */ -static void disas_extract(DisasContext *s, uint32_t insn) +static bool trans_EXTR(DisasContext *s, arg_extract *a) { - unsigned int sf, n, rm, imm, rn, rd, bitsize, op21, op0; + TCGv_i64 tcg_rd, tcg_rm, tcg_rn; - sf = extract32(insn, 31, 1); - n = extract32(insn, 22, 1); - rm = extract32(insn, 16, 5); - imm = extract32(insn, 10, 6); - rn = extract32(insn, 5, 5); - rd = extract32(insn, 0, 5); - op21 = extract32(insn, 29, 2); - op0 = extract32(insn, 21, 1); - bitsize = sf ? 64 : 32; + tcg_rd = cpu_reg(s, a->rd); - if (sf != n || op21 || op0 || imm >= bitsize) { - unallocated_encoding(s); - } else { - TCGv_i64 tcg_rd, tcg_rm, tcg_rn; - - tcg_rd = cpu_reg(s, rd); - - if (unlikely(imm == 0)) { - /* tcg shl_i32/shl_i64 is undefined for 32/64 bit shifts, - * so an extract from bit 0 is a special case. - */ - if (sf) { - tcg_gen_mov_i64(tcg_rd, cpu_reg(s, rm)); - } else { - tcg_gen_ext32u_i64(tcg_rd, cpu_reg(s, rm)); - } + if (unlikely(a->imm == 0)) { + /* + * tcg shl_i32/shl_i64 is undefined for 32/64 bit shifts, + * so an extract from bit 0 is a special case. + */ + if (a->sf) { + tcg_gen_mov_i64(tcg_rd, cpu_reg(s, a->rm)); } else { - tcg_rm = cpu_reg(s, rm); - tcg_rn = cpu_reg(s, rn); + tcg_gen_ext32u_i64(tcg_rd, cpu_reg(s, a->rm)); + } + } else { + tcg_rm = cpu_reg(s, a->rm); + tcg_rn = cpu_reg(s, a->rn); - if (sf) { - /* Specialization to ROR happens in EXTRACT2. */ - tcg_gen_extract2_i64(tcg_rd, tcg_rm, tcg_rn, imm); + if (a->sf) { + /* Specialization to ROR happens in EXTRACT2. */ + tcg_gen_extract2_i64(tcg_rd, tcg_rm, tcg_rn, a->imm); + } else { + TCGv_i32 t0 = tcg_temp_new_i32(); + + tcg_gen_extrl_i64_i32(t0, tcg_rm); + if (a->rm == a->rn) { + tcg_gen_rotri_i32(t0, t0, a->imm); } else { - TCGv_i32 t0 = tcg_temp_new_i32(); - - tcg_gen_extrl_i64_i32(t0, tcg_rm); - if (rm == rn) { - tcg_gen_rotri_i32(t0, t0, imm); - } else { - TCGv_i32 t1 = tcg_temp_new_i32(); - tcg_gen_extrl_i64_i32(t1, tcg_rn); - tcg_gen_extract2_i32(t0, t0, t1, imm); - } - tcg_gen_extu_i32_i64(tcg_rd, t0); + TCGv_i32 t1 = tcg_temp_new_i32(); + tcg_gen_extrl_i64_i32(t1, tcg_rn); + tcg_gen_extract2_i32(t0, t0, t1, a->imm); } + tcg_gen_extu_i32_i64(tcg_rd, t0); } } -} - -/* Data processing - immediate */ -static void disas_data_proc_imm(DisasContext *s, uint32_t insn) -{ - switch (extract32(insn, 23, 6)) { - case 0x27: /* Extract */ - disas_extract(s, insn); - break; - default: - unallocated_encoding(s); - break; - } + return true; } /* Shift a TCGv src by TCGv shift_amount, put result in dst. @@ -XXX,XX +XXX,XX @@ static bool btype_destination_ok(uint32_t insn, bool bt, int btype) static void disas_a64_legacy(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 4)) { - case 0x8: case 0x9: /* Data processing - immediate */ - disas_data_proc_imm(s, insn); - break; case 0xa: case 0xb: /* Branch, exception generation and system insns */ disas_b_exc_sys(s, insn); break; -- 2.34.1 Convert the unconditional branch immediate insns B and BL to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-14-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 9 +++++++++ target/arm/tcg/translate-a64.c | 31 +++++++++++-------------------- 2 files changed, 20 insertions(+), 20 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ &ri rd imm &rri_sf rd rn imm sf +&i imm ### Data Processing - Immediate @@ -XXX,XX +XXX,XX @@ UBFM . 10 100110 . ...... ...... ..... ..... @bitfield_32 EXTR 1 00 100111 1 0 rm:5 imm:6 rn:5 rd:5 &extract sf=1 EXTR 0 00 100111 0 0 rm:5 0 imm:5 rn:5 rd:5 &extract sf=0 + +# Branches + +%imm26 0:s26 !function=times_4 +@branch . ..... .......................... &i imm=%imm26 + +B 0 00101 .......................... @branch +BL 1 00101 .......................... @branch diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static inline AArch64DecodeFn *lookup_disas_fn(const AArch64DecodeTable *table, * match up with those in the manual. */ -/* Unconditional branch (immediate) - * 31 30 26 25 0 - * +----+-----------+-------------------------------------+ - * | op | 0 0 1 0 1 | imm26 | - * +----+-----------+-------------------------------------+ - */ -static void disas_uncond_b_imm(DisasContext *s, uint32_t insn) +static bool trans_B(DisasContext *s, arg_i *a) { - int64_t diff = sextract32(insn, 0, 26) * 4; - - if (insn & (1U << 31)) { - /* BL Branch with link */ - gen_pc_plus_diff(s, cpu_reg(s, 30), curr_insn_len(s)); - } - - /* B Branch / BL Branch with link */ reset_btype(s); - gen_goto_tb(s, 0, diff); + gen_goto_tb(s, 0, a->imm); + return true; +} + +static bool trans_BL(DisasContext *s, arg_i *a) +{ + gen_pc_plus_diff(s, cpu_reg(s, 30), curr_insn_len(s)); + reset_btype(s); + gen_goto_tb(s, 0, a->imm); + return true; } /* Compare and branch (immediate) @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) static void disas_b_exc_sys(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 7)) { - case 0x0a: case 0x0b: - case 0x4a: case 0x4b: /* Unconditional branch (immediate) */ - disas_uncond_b_imm(s, insn); - break; case 0x1a: case 0x5a: /* Compare & branch (immediate) */ disas_comp_b_imm(s, insn); break; -- 2.34.1 Convert the compare-and-branch-immediate insns CBZ and CBNZ to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-15-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 5 +++++ target/arm/tcg/translate-a64.c | 26 ++++++-------------------- 2 files changed, 11 insertions(+), 20 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ EXTR 0 00 100111 0 0 rm:5 0 imm:5 rn:5 rd:5 &extract sf=0 B 0 00101 .......................... @branch BL 1 00101 .......................... @branch + +%imm19 5:s19 !function=times_4 +&cbz rt imm sf nz + +CBZ sf:1 011010 nz:1 ................... rt:5 &cbz imm=%imm19 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_BL(DisasContext *s, arg_i *a) return true; } -/* Compare and branch (immediate) - * 31 30 25 24 23 5 4 0 - * +----+-------------+----+---------------------+--------+ - * | sf | 0 1 1 0 1 0 | op | imm19 | Rt | - * +----+-------------+----+---------------------+--------+ - */ -static void disas_comp_b_imm(DisasContext *s, uint32_t insn) + +static bool trans_CBZ(DisasContext *s, arg_cbz *a) { - unsigned int sf, op, rt; - int64_t diff; DisasLabel match; TCGv_i64 tcg_cmp; - sf = extract32(insn, 31, 1); - op = extract32(insn, 24, 1); /* 0: CBZ; 1: CBNZ */ - rt = extract32(insn, 0, 5); - diff = sextract32(insn, 5, 19) * 4; - - tcg_cmp = read_cpu_reg(s, rt, sf); + tcg_cmp = read_cpu_reg(s, a->rt, a->sf); reset_btype(s); match = gen_disas_label(s); - tcg_gen_brcondi_i64(op ? TCG_COND_NE : TCG_COND_EQ, + tcg_gen_brcondi_i64(a->nz ? TCG_COND_NE : TCG_COND_EQ, tcg_cmp, 0, match.label); gen_goto_tb(s, 0, 4); set_disas_label(s, match); - gen_goto_tb(s, 1, diff); + gen_goto_tb(s, 1, a->imm); + return true; } /* Test and branch (immediate) @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) static void disas_b_exc_sys(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 7)) { - case 0x1a: case 0x5a: /* Compare & branch (immediate) */ - disas_comp_b_imm(s, insn); - break; case 0x1b: case 0x5b: /* Test & branch (immediate) */ disas_test_b_imm(s, insn); break; -- 2.34.1 Convert the test-and-branch-immediate insns TBZ and TBNZ to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-16-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 6 ++++++ target/arm/tcg/translate-a64.c | 25 +++++-------------------- 2 files changed, 11 insertions(+), 20 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ BL 1 00101 .......................... @branch &cbz rt imm sf nz CBZ sf:1 011010 nz:1 ................... rt:5 &cbz imm=%imm19 + +%imm14 5:s14 !function=times_4 +%imm31_19 31:1 19:5 +&tbz rt imm nz bitpos + +TBZ . 011011 nz:1 ..... .............. rt:5 &tbz imm=%imm14 bitpos=%imm31_19 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_CBZ(DisasContext *s, arg_cbz *a) return true; } -/* Test and branch (immediate) - * 31 30 25 24 23 19 18 5 4 0 - * +----+-------------+----+-------+-------------+------+ - * | b5 | 0 1 1 0 1 1 | op | b40 | imm14 | Rt | - * +----+-------------+----+-------+-------------+------+ - */ -static void disas_test_b_imm(DisasContext *s, uint32_t insn) +static bool trans_TBZ(DisasContext *s, arg_tbz *a) { - unsigned int bit_pos, op, rt; - int64_t diff; DisasLabel match; TCGv_i64 tcg_cmp; - bit_pos = (extract32(insn, 31, 1) << 5) | extract32(insn, 19, 5); - op = extract32(insn, 24, 1); /* 0: TBZ; 1: TBNZ */ - diff = sextract32(insn, 5, 14) * 4; - rt = extract32(insn, 0, 5); - tcg_cmp = tcg_temp_new_i64(); - tcg_gen_andi_i64(tcg_cmp, cpu_reg(s, rt), (1ULL << bit_pos)); + tcg_gen_andi_i64(tcg_cmp, cpu_reg(s, a->rt), 1ULL << a->bitpos); reset_btype(s); match = gen_disas_label(s); - tcg_gen_brcondi_i64(op ? TCG_COND_NE : TCG_COND_EQ, + tcg_gen_brcondi_i64(a->nz ? TCG_COND_NE : TCG_COND_EQ, tcg_cmp, 0, match.label); gen_goto_tb(s, 0, 4); set_disas_label(s, match); - gen_goto_tb(s, 1, diff); + gen_goto_tb(s, 1, a->imm); + return true; } /* Conditional branch (immediate) @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) static void disas_b_exc_sys(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 7)) { - case 0x1b: case 0x5b: /* Test & branch (immediate) */ - disas_test_b_imm(s, insn); - break; case 0x2a: /* Conditional branch (immediate) */ disas_cond_b_imm(s, insn); break; -- 2.34.1 Convert the immediate conditional branch insn B.cond to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-17-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 2 ++ target/arm/tcg/translate-a64.c | 30 ++++++------------------------ 2 files changed, 8 insertions(+), 24 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ CBZ sf:1 011010 nz:1 ................... rt:5 &cbz imm=%imm19 &tbz rt imm nz bitpos TBZ . 011011 nz:1 ..... .............. rt:5 &tbz imm=%imm14 bitpos=%imm31_19 + +B_cond 0101010 0 ................... 0 cond:4 imm=%imm19 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_TBZ(DisasContext *s, arg_tbz *a) return true; } -/* Conditional branch (immediate) - * 31 25 24 23 5 4 3 0 - * +---------------+----+---------------------+----+------+ - * | 0 1 0 1 0 1 0 | o1 | imm19 | o0 | cond | - * +---------------+----+---------------------+----+------+ - */ -static void disas_cond_b_imm(DisasContext *s, uint32_t insn) +static bool trans_B_cond(DisasContext *s, arg_B_cond *a) { - unsigned int cond; - int64_t diff; - - if ((insn & (1 << 4)) || (insn & (1 << 24))) { - unallocated_encoding(s); - return; - } - diff = sextract32(insn, 5, 19) * 4; - cond = extract32(insn, 0, 4); - reset_btype(s); - if (cond < 0x0e) { + if (a->cond < 0x0e) { /* genuinely conditional branches */ DisasLabel match = gen_disas_label(s); - arm_gen_test_cc(cond, match.label); + arm_gen_test_cc(a->cond, match.label); gen_goto_tb(s, 0, 4); set_disas_label(s, match); - gen_goto_tb(s, 1, diff); + gen_goto_tb(s, 1, a->imm); } else { /* 0xe and 0xf are both "always" conditions */ - gen_goto_tb(s, 0, diff); + gen_goto_tb(s, 0, a->imm); } + return true; } /* HINT instruction group, including various allocated HINTs */ @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) static void disas_b_exc_sys(DisasContext *s, uint32_t insn) { switch (extract32(insn, 25, 7)) { - case 0x2a: /* Conditional branch (immediate) */ - disas_cond_b_imm(s, insn); - break; case 0x6a: /* Exception generation / System */ if (insn & (1 << 24)) { if (extract32(insn, 22, 2) == 0) { -- 2.34.1 Convert the simple (non-pointer-auth) BR, BLR and RET insns to decodetree. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-18-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 5 ++++ target/arm/tcg/translate-a64.c | 55 ++++++++++++++++++++++++++++++---- 2 files changed, 54 insertions(+), 6 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ # This file is processed by scripts/decodetree.py # +&r rn &ri rd imm &rri_sf rd rn imm sf &i imm @@ -XXX,XX +XXX,XX @@ CBZ sf:1 011010 nz:1 ................... rt:5 &cbz imm=%imm19 TBZ . 011011 nz:1 ..... .............. rt:5 &tbz imm=%imm14 bitpos=%imm31_19 B_cond 0101010 0 ................... 0 cond:4 imm=%imm19 + +BR 1101011 0000 11111 000000 rn:5 00000 &r +BLR 1101011 0001 11111 000000 rn:5 00000 &r +RET 1101011 0010 11111 000000 rn:5 00000 &r diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_B_cond(DisasContext *s, arg_B_cond *a) return true; } +static void set_btype_for_br(DisasContext *s, int rn) +{ + if (dc_isar_feature(aa64_bti, s)) { + /* BR to {x16,x17} or !guard -> 1, else 3. */ + set_btype(s, rn == 16 || rn == 17 || !s->guarded_page ? 1 : 3); + } +} + +static void set_btype_for_blr(DisasContext *s) +{ + if (dc_isar_feature(aa64_bti, s)) { + /* BLR sets BTYPE to 2, regardless of source guarded page. */ + set_btype(s, 2); + } +} + +static bool trans_BR(DisasContext *s, arg_r *a) +{ + gen_a64_set_pc(s, cpu_reg(s, a->rn)); + set_btype_for_br(s, a->rn); + s->base.is_jmp = DISAS_JUMP; + return true; +} + +static bool trans_BLR(DisasContext *s, arg_r *a) +{ + TCGv_i64 dst = cpu_reg(s, a->rn); + TCGv_i64 lr = cpu_reg(s, 30); + if (dst == lr) { + TCGv_i64 tmp = tcg_temp_new_i64(); + tcg_gen_mov_i64(tmp, dst); + dst = tmp; + } + gen_pc_plus_diff(s, lr, curr_insn_len(s)); + gen_a64_set_pc(s, dst); + set_btype_for_blr(s); + s->base.is_jmp = DISAS_JUMP; + return true; +} + +static bool trans_RET(DisasContext *s, arg_r *a) +{ + gen_a64_set_pc(s, cpu_reg(s, a->rn)); + s->base.is_jmp = DISAS_JUMP; + return true; +} + /* HINT instruction group, including various allocated HINTs */ static void handle_hint(DisasContext *s, uint32_t insn, unsigned int op1, unsigned int op2, unsigned int crm) @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) btype_mod = opc; switch (op3) { case 0: - /* BR, BLR, RET */ - if (op4 != 0) { - goto do_unallocated; - } - dst = cpu_reg(s, rn); - break; + /* BR, BLR, RET : handled in decodetree */ + goto do_unallocated; case 2: case 3: -- 2.34.1 Convert the single-register pointer-authentication variants of BR, BLR, RET to decodetree. (BRAA/BLRAA are in a different branch of the legacy decoder and will be dealt with in the next commit.) Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-19-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 7 ++ target/arm/tcg/translate-a64.c | 132 +++++++++++++++++++-------------- 2 files changed, 84 insertions(+), 55 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ B_cond 0101010 0 ................... 0 cond:4 imm=%imm19 BR 1101011 0000 11111 000000 rn:5 00000 &r BLR 1101011 0001 11111 000000 rn:5 00000 &r RET 1101011 0010 11111 000000 rn:5 00000 &r + +&braz rn m +BRAZ 1101011 0000 11111 00001 m:1 rn:5 11111 &braz # BRAAZ, BRABZ +BLRAZ 1101011 0001 11111 00001 m:1 rn:5 11111 &braz # BLRAAZ, BLRABZ + +&reta m +RETA 1101011 0010 11111 00001 m:1 11111 11111 &reta # RETAA, RETAB diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_RET(DisasContext *s, arg_r *a) return true; } +static TCGv_i64 auth_branch_target(DisasContext *s, TCGv_i64 dst, + TCGv_i64 modifier, bool use_key_a) +{ + TCGv_i64 truedst; + /* + * Return the branch target for a BRAA/RETA/etc, which is either + * just the destination dst, or that value with the pauth check + * done and the code removed from the high bits. + */ + if (!s->pauth_active) { + return dst; + } + + truedst = tcg_temp_new_i64(); + if (use_key_a) { + gen_helper_autia(truedst, cpu_env, dst, modifier); + } else { + gen_helper_autib(truedst, cpu_env, dst, modifier); + } + return truedst; +} + +static bool trans_BRAZ(DisasContext *s, arg_braz *a) +{ + TCGv_i64 dst; + + if (!dc_isar_feature(aa64_pauth, s)) { + return false; + } + + dst = auth_branch_target(s, cpu_reg(s, a->rn), tcg_constant_i64(0), !a->m); + gen_a64_set_pc(s, dst); + set_btype_for_br(s, a->rn); + s->base.is_jmp = DISAS_JUMP; + return true; +} + +static bool trans_BLRAZ(DisasContext *s, arg_braz *a) +{ + TCGv_i64 dst, lr; + + if (!dc_isar_feature(aa64_pauth, s)) { + return false; + } + + dst = auth_branch_target(s, cpu_reg(s, a->rn), tcg_constant_i64(0), !a->m); + lr = cpu_reg(s, 30); + if (dst == lr) { + TCGv_i64 tmp = tcg_temp_new_i64(); + tcg_gen_mov_i64(tmp, dst); + dst = tmp; + } + gen_pc_plus_diff(s, lr, curr_insn_len(s)); + gen_a64_set_pc(s, dst); + set_btype_for_blr(s); + s->base.is_jmp = DISAS_JUMP; + return true; +} + +static bool trans_RETA(DisasContext *s, arg_reta *a) +{ + TCGv_i64 dst; + + dst = auth_branch_target(s, cpu_reg(s, 30), cpu_X[31], !a->m); + gen_a64_set_pc(s, dst); + s->base.is_jmp = DISAS_JUMP; + return true; +} + /* HINT instruction group, including various allocated HINTs */ static void handle_hint(DisasContext *s, uint32_t insn, unsigned int op1, unsigned int op2, unsigned int crm) @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) } switch (opc) { - case 0: /* BR */ - case 1: /* BLR */ - case 2: /* RET */ - btype_mod = opc; - switch (op3) { - case 0: - /* BR, BLR, RET : handled in decodetree */ - goto do_unallocated; - - case 2: - case 3: - if (!dc_isar_feature(aa64_pauth, s)) { - goto do_unallocated; - } - if (opc == 2) { - /* RETAA, RETAB */ - if (rn != 0x1f || op4 != 0x1f) { - goto do_unallocated; - } - rn = 30; - modifier = cpu_X[31]; - } else { - /* BRAAZ, BRABZ, BLRAAZ, BLRABZ */ - if (op4 != 0x1f) { - goto do_unallocated; - } - modifier = tcg_constant_i64(0); - } - if (s->pauth_active) { - dst = tcg_temp_new_i64(); - if (op3 == 2) { - gen_helper_autia(dst, cpu_env, cpu_reg(s, rn), modifier); - } else { - gen_helper_autib(dst, cpu_env, cpu_reg(s, rn), modifier); - } - } else { - dst = cpu_reg(s, rn); - } - break; - - default: - goto do_unallocated; - } - /* BLR also needs to load return address */ - if (opc == 1) { - TCGv_i64 lr = cpu_reg(s, 30); - if (dst == lr) { - TCGv_i64 tmp = tcg_temp_new_i64(); - tcg_gen_mov_i64(tmp, dst); - dst = tmp; - } - gen_pc_plus_diff(s, lr, curr_insn_len(s)); - } - gen_a64_set_pc(s, dst); - break; + case 0: + case 1: + case 2: + /* + * BR, BLR, RET, RETAA, RETAB, BRAAZ, BRABZ, BLRAAZ, BLRABZ: + * handled in decodetree + */ + goto do_unallocated; case 8: /* BRAA */ case 9: /* BLRAA */ -- 2.34.1 Convert the last four BR-with-pointer-auth insns to decodetree. The remaining cases in the outer switch in disas_uncond_b_reg() all return early rather than leaving the case statement, so we can delete the now-unused code at the end of that function. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-20-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 4 ++ target/arm/tcg/translate-a64.c | 97 ++++++++++++++-------------------- 2 files changed, 43 insertions(+), 58 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ BLRAZ 1101011 0001 11111 00001 m:1 rn:5 11111 &braz # BLRAAZ, BLRABZ &reta m RETA 1101011 0010 11111 00001 m:1 11111 11111 &reta # RETAA, RETAB + +&bra rn rm m +BRA 1101011 1000 11111 00001 m:1 rn:5 rm:5 &bra # BRAA, BRAB +BLRA 1101011 1001 11111 00001 m:1 rn:5 rm:5 &bra # BLRAA, BLRAB diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_RETA(DisasContext *s, arg_reta *a) return true; } +static bool trans_BRA(DisasContext *s, arg_bra *a) +{ + TCGv_i64 dst; + + if (!dc_isar_feature(aa64_pauth, s)) { + return false; + } + dst = auth_branch_target(s, cpu_reg(s,a->rn), cpu_reg_sp(s, a->rm), !a->m); + gen_a64_set_pc(s, dst); + set_btype_for_br(s, a->rn); + s->base.is_jmp = DISAS_JUMP; + return true; +} + +static bool trans_BLRA(DisasContext *s, arg_bra *a) +{ + TCGv_i64 dst, lr; + + if (!dc_isar_feature(aa64_pauth, s)) { + return false; + } + dst = auth_branch_target(s, cpu_reg(s, a->rn), cpu_reg_sp(s, a->rm), !a->m); + lr = cpu_reg(s, 30); + if (dst == lr) { + TCGv_i64 tmp = tcg_temp_new_i64(); + tcg_gen_mov_i64(tmp, dst); + dst = tmp; + } + gen_pc_plus_diff(s, lr, curr_insn_len(s)); + gen_a64_set_pc(s, dst); + set_btype_for_blr(s); + s->base.is_jmp = DISAS_JUMP; + return true; +} + /* HINT instruction group, including various allocated HINTs */ static void handle_hint(DisasContext *s, uint32_t insn, unsigned int op1, unsigned int op2, unsigned int crm) @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) { unsigned int opc, op2, op3, rn, op4; - unsigned btype_mod = 2; /* 0: BR, 1: BLR, 2: other */ TCGv_i64 dst; TCGv_i64 modifier; @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) case 0: case 1: case 2: + case 8: + case 9: /* - * BR, BLR, RET, RETAA, RETAB, BRAAZ, BRABZ, BLRAAZ, BLRABZ: - * handled in decodetree + * BR, BLR, RET, RETAA, RETAB, BRAAZ, BRABZ, BLRAAZ, BLRABZ, + * BRAA, BLRAA: handled in decodetree */ goto do_unallocated; - case 8: /* BRAA */ - case 9: /* BLRAA */ - if (!dc_isar_feature(aa64_pauth, s)) { - goto do_unallocated; - } - if ((op3 & ~1) != 2) { - goto do_unallocated; - } - btype_mod = opc & 1; - if (s->pauth_active) { - dst = tcg_temp_new_i64(); - modifier = cpu_reg_sp(s, op4); - if (op3 == 2) { - gen_helper_autia(dst, cpu_env, cpu_reg(s, rn), modifier); - } else { - gen_helper_autib(dst, cpu_env, cpu_reg(s, rn), modifier); - } - } else { - dst = cpu_reg(s, rn); - } - /* BLRAA also needs to load return address */ - if (opc == 9) { - TCGv_i64 lr = cpu_reg(s, 30); - if (dst == lr) { - TCGv_i64 tmp = tcg_temp_new_i64(); - tcg_gen_mov_i64(tmp, dst); - dst = tmp; - } - gen_pc_plus_diff(s, lr, curr_insn_len(s)); - } - gen_a64_set_pc(s, dst); - break; - case 4: /* ERET */ if (s->current_el == 0) { goto do_unallocated; @@ -XXX,XX +XXX,XX @@ static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) unallocated_encoding(s); return; } - - switch (btype_mod) { - case 0: /* BR */ - if (dc_isar_feature(aa64_bti, s)) { - /* BR to {x16,x17} or !guard -> 1, else 3. */ - set_btype(s, rn == 16 || rn == 17 || !s->guarded_page ? 1 : 3); - } - break; - - case 1: /* BLR */ - if (dc_isar_feature(aa64_bti, s)) { - /* BLR sets BTYPE to 2, regardless of source guarded page. */ - set_btype(s, 2); - } - break; - - default: /* RET or none of the above. */ - /* BTYPE will be set to 0 by normal end-of-insn processing. */ - break; - } - - s->base.is_jmp = DISAS_JUMP; } /* Branches, exception generating and system instructions */ -- 2.34.1 Convert the exception-return insns ERET, ERETA and ERETB to decodetree. These were the last insns left in the legacy decoder function disas_uncond_reg_b(), which allows us to remove it. The old decoder explicitly decoded the DRPS instruction, only in order to call unallocated_encoding() on it, exactly as would have happened if it hadn't decoded it. This is because this insn always UNDEFs unless the CPU is in halting-debug state, which we don't emulate. So we list the pattern in a comment in a64.decode, but don't actively decode it. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512144106.3608981-21-peter.maydell@linaro.org --- target/arm/tcg/a64.decode | 8 ++ target/arm/tcg/translate-a64.c | 163 +++++++++++---------------------- 2 files changed, 63 insertions(+), 108 deletions(-) diff --git a/target/arm/tcg/a64.decode b/target/arm/tcg/a64.decode index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/a64.decode +++ b/target/arm/tcg/a64.decode @@ -XXX,XX +XXX,XX @@ RETA 1101011 0010 11111 00001 m:1 11111 11111 &reta # RETAA, RETAB &bra rn rm m BRA 1101011 1000 11111 00001 m:1 rn:5 rm:5 &bra # BRAA, BRAB BLRA 1101011 1001 11111 00001 m:1 rn:5 rm:5 &bra # BLRAA, BLRAB + +ERET 1101011 0100 11111 000000 11111 00000 +ERETA 1101011 0100 11111 00001 m:1 11111 11111 &reta # ERETAA, ERETAB + +# We don't need to decode DRPS because it always UNDEFs except when +# the processor is in halting debug state (which we don't implement). +# The pattern is listed here as documentation. +# DRPS 1101011 0101 11111 000000 11111 00000 diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/tcg/translate-a64.c +++ b/target/arm/tcg/translate-a64.c @@ -XXX,XX +XXX,XX @@ static bool trans_BLRA(DisasContext *s, arg_bra *a) return true; } +static bool trans_ERET(DisasContext *s, arg_ERET *a) +{ + TCGv_i64 dst; + + if (s->current_el == 0) { + return false; + } + if (s->fgt_eret) { + gen_exception_insn_el(s, 0, EXCP_UDEF, 0, 2); + return true; + } + dst = tcg_temp_new_i64(); + tcg_gen_ld_i64(dst, cpu_env, + offsetof(CPUARMState, elr_el[s->current_el])); + + if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { + gen_io_start(); + } + + gen_helper_exception_return(cpu_env, dst); + /* Must exit loop to check un-masked IRQs */ + s->base.is_jmp = DISAS_EXIT; + return true; +} + +static bool trans_ERETA(DisasContext *s, arg_reta *a) +{ + TCGv_i64 dst; + + if (!dc_isar_feature(aa64_pauth, s)) { + return false; + } + if (s->current_el == 0) { + return false; + } + /* The FGT trap takes precedence over an auth trap. */ + if (s->fgt_eret) { + gen_exception_insn_el(s, 0, EXCP_UDEF, a->m ? 3 : 2, 2); + return true; + } + dst = tcg_temp_new_i64(); + tcg_gen_ld_i64(dst, cpu_env, + offsetof(CPUARMState, elr_el[s->current_el])); + + dst = auth_branch_target(s, dst, cpu_X[31], !a->m); + if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { + gen_io_start(); + } + + gen_helper_exception_return(cpu_env, dst); + /* Must exit loop to check un-masked IRQs */ + s->base.is_jmp = DISAS_EXIT; + return true; +} + /* HINT instruction group, including various allocated HINTs */ static void handle_hint(DisasContext *s, uint32_t insn, unsigned int op1, unsigned int op2, unsigned int crm) @@ -XXX,XX +XXX,XX @@ static void disas_exc(DisasContext *s, uint32_t insn) } } -/* Unconditional branch (register) - * 31 25 24 21 20 16 15 10 9 5 4 0 - * +---------------+-------+-------+-------+------+-------+ - * | 1 1 0 1 0 1 1 | opc | op2 | op3 | Rn | op4 | - * +---------------+-------+-------+-------+------+-------+ - */ -static void disas_uncond_b_reg(DisasContext *s, uint32_t insn) -{ - unsigned int opc, op2, op3, rn, op4; - TCGv_i64 dst; - TCGv_i64 modifier; - - opc = extract32(insn, 21, 4); - op2 = extract32(insn, 16, 5); - op3 = extract32(insn, 10, 6); - rn = extract32(insn, 5, 5); - op4 = extract32(insn, 0, 5); - - if (op2 != 0x1f) { - goto do_unallocated; - } - - switch (opc) { - case 0: - case 1: - case 2: - case 8: - case 9: - /* - * BR, BLR, RET, RETAA, RETAB, BRAAZ, BRABZ, BLRAAZ, BLRABZ, - * BRAA, BLRAA: handled in decodetree - */ - goto do_unallocated; - - case 4: /* ERET */ - if (s->current_el == 0) { - goto do_unallocated; - } - switch (op3) { - case 0: /* ERET */ - if (op4 != 0) { - goto do_unallocated; - } - if (s->fgt_eret) { - gen_exception_insn_el(s, 0, EXCP_UDEF, syn_erettrap(op3), 2); - return; - } - dst = tcg_temp_new_i64(); - tcg_gen_ld_i64(dst, cpu_env, - offsetof(CPUARMState, elr_el[s->current_el])); - break; - - case 2: /* ERETAA */ - case 3: /* ERETAB */ - if (!dc_isar_feature(aa64_pauth, s)) { - goto do_unallocated; - } - if (rn != 0x1f || op4 != 0x1f) { - goto do_unallocated; - } - /* The FGT trap takes precedence over an auth trap. */ - if (s->fgt_eret) { - gen_exception_insn_el(s, 0, EXCP_UDEF, syn_erettrap(op3), 2); - return; - } - dst = tcg_temp_new_i64(); - tcg_gen_ld_i64(dst, cpu_env, - offsetof(CPUARMState, elr_el[s->current_el])); - if (s->pauth_active) { - modifier = cpu_X[31]; - if (op3 == 2) { - gen_helper_autia(dst, cpu_env, dst, modifier); - } else { - gen_helper_autib(dst, cpu_env, dst, modifier); - } - } - break; - - default: - goto do_unallocated; - } - if (tb_cflags(s->base.tb) & CF_USE_ICOUNT) { - gen_io_start(); - } - - gen_helper_exception_return(cpu_env, dst); - /* Must exit loop to check un-masked IRQs */ - s->base.is_jmp = DISAS_EXIT; - return; - - case 5: /* DRPS */ - if (op3 != 0 || op4 != 0 || rn != 0x1f) { - goto do_unallocated; - } else { - unallocated_encoding(s); - } - return; - - default: - do_unallocated: - unallocated_encoding(s); - return; - } -} - /* Branches, exception generating and system instructions */ static void disas_b_exc_sys(DisasContext *s, uint32_t insn) { @@ -XXX,XX +XXX,XX @@ static void disas_b_exc_sys(DisasContext *s, uint32_t insn) disas_exc(s, insn); } break; - case 0x6b: /* Unconditional branch (register) */ - disas_uncond_b_reg(s, insn); - break; default: unallocated_encoding(s); break; -- 2.34.1 The IMPDEF sysreg L2CTLR_EL1 found on the Cortex-A35, A53, A57, A72 and which we (arguably dubiously) also provide in '-cpu max' has a 2 bit field for the number of processors in the cluster. On real hardware this must be sufficient because it can only be configured with up to 4 CPUs in the cluster. However on QEMU if the board code does not explicitly configure the code into clusters with the right CPU count we default to "give the value assuming that all CPUs in the system are in a single cluster", which might be too big to fit in the field. Instead of just overflowing this 2-bit field, saturate to 3 (meaning "4 CPUs", so at least we don't overwrite other fields in the register. It's unlikely that any guest code really cares about the value in this field; at least, if it does it probably also wants the system to be more closely matching real hardware, i.e. not to have more than 4 CPUs. This issue has been present since the L2CTLR was first added in commit 377a44ec8f2fac5b back in 2014. It was only noticed because Coverity complains (CID 1509227) that the shift might overflow 32 bits and inadvertently sign extend into the top half of the 64 bit value. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Message-id: 20230512170223.3801643-2-peter.maydell@linaro.org --- target/arm/cortex-regs.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/target/arm/cortex-regs.c b/target/arm/cortex-regs.c index XXXXXXX..XXXXXXX 100644 --- a/target/arm/cortex-regs.c +++ b/target/arm/cortex-regs.c @@ -XXX,XX +XXX,XX @@ static uint64_t l2ctlr_read(CPUARMState *env, const ARMCPRegInfo *ri) { ARMCPU *cpu = env_archcpu(env); - /* Number of cores is in [25:24]; otherwise we RAZ */ - return (cpu->core_count - 1) << 24; + /* + * Number of cores is in [25:24]; otherwise we RAZ. + * If the board didn't configure the CPUs into clusters, + * we default to "all CPUs in one cluster", which might be + * more than the 4 that the hardware permits and which is + * all you can report in this two-bit field. Saturate to + * 0b11 (== 4 CPUs) rather than overflowing the field. + */ + return MIN(cpu->core_count - 1, 3) << 24; } static const ARMCPRegInfo cortex_a72_a57_a53_cp_reginfo[] = { -- 2.34.1 In the vexpress board code, we allocate a new MemoryRegion at the top of vexpress_common_init() but only set it up and use it inside the "if (map[VE_NORFLASHALIAS] != -1)" conditional, so we leak it if not. This isn't a very interesting leak as it's a tiny amount of memory once at startup, but it's easy to fix. We could silence Coverity simply by moving the g_new() into the if() block, but this use of g_new(MemoryRegion, 1) is a legacy from when this board model was originally written; we wouldn't do that if we wrote it today. The MemoryRegions are conceptually a part of the board and must not go away until the whole board is done with (at the end of the simulation), so they belong in its state struct. This machine already has a VexpressMachineState struct that extends MachineState, so statically put the MemoryRegions in there instead of dynamically allocating them separately at runtime. Spotted by Coverity (CID 1509083). Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org> Message-id: 20230512170223.3801643-3-peter.maydell@linaro.org --- hw/arm/vexpress.c | 40 ++++++++++++++++++++-------------------- 1 file changed, 20 insertions(+), 20 deletions(-) diff --git a/hw/arm/vexpress.c b/hw/arm/vexpress.c index XXXXXXX..XXXXXXX 100644 --- a/hw/arm/vexpress.c +++ b/hw/arm/vexpress.c @@ -XXX,XX +XXX,XX @@ struct VexpressMachineClass { struct VexpressMachineState { MachineState parent; + MemoryRegion vram; + MemoryRegion sram; + MemoryRegion flashalias; + MemoryRegion lowram; + MemoryRegion a15sram; bool secure; bool virt; }; @@ -XXX,XX +XXX,XX @@ struct VexpressMachineState { #define TYPE_VEXPRESS_A15_MACHINE MACHINE_TYPE_NAME("vexpress-a15") OBJECT_DECLARE_TYPE(VexpressMachineState, VexpressMachineClass, VEXPRESS_MACHINE) -typedef void DBoardInitFn(const VexpressMachineState *machine, +typedef void DBoardInitFn(VexpressMachineState *machine, ram_addr_t ram_size, const char *cpu_type, qemu_irq *pic); @@ -XXX,XX +XXX,XX @@ static void init_cpus(MachineState *ms, const char *cpu_type, } } -static void a9_daughterboard_init(const VexpressMachineState *vms, +static void a9_daughterboard_init(VexpressMachineState *vms, ram_addr_t ram_size, const char *cpu_type, qemu_irq *pic) { MachineState *machine = MACHINE(vms); MemoryRegion *sysmem = get_system_memory(); - MemoryRegion *lowram = g_new(MemoryRegion, 1); ram_addr_t low_ram_size; if (ram_size > 0x40000000) { @@ -XXX,XX +XXX,XX @@ static void a9_daughterboard_init(const VexpressMachineState *vms, * address space should in theory be remappable to various * things including ROM or RAM; we always map the RAM there. */ - memory_region_init_alias(lowram, NULL, "vexpress.lowmem", machine->ram, - 0, low_ram_size); - memory_region_add_subregion(sysmem, 0x0, lowram); + memory_region_init_alias(&vms->lowram, NULL, "vexpress.lowmem", + machine->ram, 0, low_ram_size); + memory_region_add_subregion(sysmem, 0x0, &vms->lowram); memory_region_add_subregion(sysmem, 0x60000000, machine->ram); /* 0x1e000000 A9MPCore (SCU) private memory region */ @@ -XXX,XX +XXX,XX @@ static VEDBoardInfo a9_daughterboard = { .init = a9_daughterboard_init, }; -static void a15_daughterboard_init(const VexpressMachineState *vms, +static void a15_daughterboard_init(VexpressMachineState *vms, ram_addr_t ram_size, const char *cpu_type, qemu_irq *pic) { MachineState *machine = MACHINE(vms); MemoryRegion *sysmem = get_system_memory(); - MemoryRegion *sram = g_new(MemoryRegion, 1); { /* We have to use a separate 64 bit variable here to avoid the gcc @@ -XXX,XX +XXX,XX @@ static void a15_daughterboard_init(const VexpressMachineState *vms, /* 0x2b060000: SP805 watchdog: not modelled */ /* 0x2b0a0000: PL341 dynamic memory controller: not modelled */ /* 0x2e000000: system SRAM */ - memory_region_init_ram(sram, NULL, "vexpress.a15sram", 0x10000, + memory_region_init_ram(&vms->a15sram, NULL, "vexpress.a15sram", 0x10000, &error_fatal); - memory_region_add_subregion(sysmem, 0x2e000000, sram); + memory_region_add_subregion(sysmem, 0x2e000000, &vms->a15sram); /* 0x7ffb0000: DMA330 DMA controller: not modelled */ /* 0x7ffd0000: PL354 static memory controller: not modelled */ @@ -XXX,XX +XXX,XX @@ static void vexpress_common_init(MachineState *machine) I2CBus *i2c; ram_addr_t vram_size, sram_size; MemoryRegion *sysmem = get_system_memory(); - MemoryRegion *vram = g_new(MemoryRegion, 1); - MemoryRegion *sram = g_new(MemoryRegion, 1); - MemoryRegion *flashalias = g_new(MemoryRegion, 1); - MemoryRegion *flash0mem; const hwaddr *map = daughterboard->motherboard_map; int i; @@ -XXX,XX +XXX,XX @@ static void vexpress_common_init(MachineState *machine) if (map[VE_NORFLASHALIAS] != -1) { /* Map flash 0 as an alias into low memory */ + MemoryRegion *flash0mem; flash0mem = sysbus_mmio_get_region(SYS_BUS_DEVICE(pflash0), 0); - memory_region_init_alias(flashalias, NULL, "vexpress.flashalias", + memory_region_init_alias(&vms->flashalias, NULL, "vexpress.flashalias", flash0mem, 0, VEXPRESS_FLASH_SIZE); - memory_region_add_subregion(sysmem, map[VE_NORFLASHALIAS], flashalias); + memory_region_add_subregion(sysmem, map[VE_NORFLASHALIAS], &vms->flashalias); } dinfo = drive_get(IF_PFLASH, 0, 1); ve_pflash_cfi01_register(map[VE_NORFLASH1], "vexpress.flash1", dinfo); sram_size = 0x2000000; - memory_region_init_ram(sram, NULL, "vexpress.sram", sram_size, + memory_region_init_ram(&vms->sram, NULL, "vexpress.sram", sram_size, &error_fatal); - memory_region_add_subregion(sysmem, map[VE_SRAM], sram); + memory_region_add_subregion(sysmem, map[VE_SRAM], &vms->sram); vram_size = 0x800000; - memory_region_init_ram(vram, NULL, "vexpress.vram", vram_size, + memory_region_init_ram(&vms->vram, NULL, "vexpress.vram", vram_size, &error_fatal); - memory_region_add_subregion(sysmem, map[VE_VIDEORAM], vram); + memory_region_add_subregion(sysmem, map[VE_VIDEORAM], &vms->vram); /* 0x4e000000 LAN9118 Ethernet */ if (nd_table[0].used) { -- 2.34.1 Convert the u2f.txt file to rST, and place it in the right place in our manual layout. The old text didn't fit very well into our manual style, so the new version ends up looking like a rewrite, although some of the original text is preserved: * the 'building' section of the old file is removed, since we generally assume that users have already built QEMU * some rather verbose text has been cut back * document the passthrough device first, on the assumption that's most likely to be of interest to users * cut back on the duplication of text between sections * format example command lines etc with rST As it's a short document it seemed simplest to do this all in one go rather than try to do a minimal syntactic conversion and then clean up the wording and layout. Signed-off-by: Peter Maydell <peter.maydell@linaro.org> Reviewed-by: Thomas Huth <thuth@redhat.com> Message-id: 20230421163734.1152076-1-peter.maydell@linaro.org --- docs/system/device-emulation.rst | 1 + docs/system/devices/usb-u2f.rst | 93 ++++++++++++++++++++++++++ docs/system/devices/usb.rst | 2 +- docs/u2f.txt | 110 ------------------------------- 4 files changed, 95 insertions(+), 111 deletions(-) create mode 100644 docs/system/devices/usb-u2f.rst delete mode 100644 docs/u2f.txt diff --git a/docs/system/device-emulation.rst b/docs/system/device-emulation.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/system/device-emulation.rst +++ b/docs/system/device-emulation.rst @@ -XXX,XX +XXX,XX @@ Emulated Devices devices/virtio-pmem.rst devices/vhost-user-rng.rst devices/canokey.rst + devices/usb-u2f.rst devices/igb.rst diff --git a/docs/system/devices/usb-u2f.rst b/docs/system/devices/usb-u2f.rst new file mode 100644 index XXXXXXX..XXXXXXX --- /dev/null +++ b/docs/system/devices/usb-u2f.rst @@ -XXX,XX +XXX,XX @@ +Universal Second Factor (U2F) USB Key Device +============================================ + +U2F is an open authentication standard that enables relying parties +exposed to the internet to offer a strong second factor option for end +user authentication. + +The second factor is provided by a device implementing the U2F +protocol. In case of a USB U2F security key, it is a USB HID device +that implements the U2F protocol. + +QEMU supports both pass-through of a host U2F key device to a VM, +and software emulation of a U2F key. + +``u2f-passthru`` +---------------- + +The ``u2f-passthru`` device allows you to connect a real hardware +U2F key on your host to a guest VM. All requests made from the guest +are passed through to the physical security key connected to the +host machine and vice versa. + +In addition, the dedicated pass-through allows you to share a single +U2F security key with several guest VMs, which is not possible with a +simple host device assignment pass-through. + +You can specify the host U2F key to use with the ``hidraw`` +option, which takes the host path to a Linux ``/dev/hidrawN`` device: + +.. parsed-literal:: + |qemu_system| -usb -device u2f-passthru,hidraw=/dev/hidraw0 + +If you don't specify the device, the ``u2f-passthru`` device will +autoscan to take the first U2F device it finds on the host (this +requires a working libudev): + +.. parsed-literal:: + |qemu_system| -usb -device u2f-passthru + +``u2f-emulated`` +---------------- + +``u2f-emulated`` is a completely software emulated U2F device. +It uses `libu2f-emu <https://github.com/MattGorko/libu2f-emu>`__ +for the U2F key emulation. libu2f-emu +provides a complete implementation of the U2F protocol device part for +all specified transports given by the FIDO Alliance. + +To work, an emulated U2F device must have four elements: + + * ec x509 certificate + * ec private key + * counter (four bytes value) + * 48 bytes of entropy (random bits) + +To use this type of device, these have to be configured, and these +four elements must be passed one way or another. + +Assuming that you have a working libu2f-emu installed on the host, +there are three possible ways to configure the ``u2f-emulated`` device: + + * ephemeral + * setup directory + * manual + +Ephemeral is the simplest way to configure; it lets the device generate +all the elements it needs for a single use of the lifetime of the device. +It is the default if you do not pass any other options to the device. + +.. parsed-literal:: + |qemu_system| -usb -device u2f-emulated + +You can pass the device the path of a setup directory on the host +using the ``dir`` option; the directory must contain these four files: + + * ``certificate.pem``: ec x509 certificate + * ``private-key.pem``: ec private key + * ``counter``: counter value + * ``entropy``: 48 bytes of entropy + +.. parsed-literal:: + |qemu_system| -usb -device u2f-emulated,dir=$dir + +You can also manually pass the device the paths to each of these files, +if you don't want them all to be in the same directory, using the options + + * ``cert`` + * ``priv`` + * ``counter`` + * ``entropy`` + +.. parsed-literal:: + |qemu_system| -usb -device u2f-emulated,cert=$DIR1/$FILE1,priv=$DIR2/$FILE2,counter=$DIR3/$FILE3,entropy=$DIR4/$FILE4 diff --git a/docs/system/devices/usb.rst b/docs/system/devices/usb.rst index XXXXXXX..XXXXXXX 100644 --- a/docs/system/devices/usb.rst +++ b/docs/system/devices/usb.rst @@ -XXX,XX +XXX,XX @@ option or the ``device_add`` monitor command. Available devices are: USB audio device ``u2f-{emulated,passthru}`` - Universal Second Factor device + :doc:`usb-u2f` ``canokey`` An Open-source Secure Key implementing FIDO2, OpenPGP, PIV and more. diff --git a/docs/u2f.txt b/docs/u2f.txt deleted file mode 100644 index XXXXXXX..XXXXXXX --- a/docs/u2f.txt +++ /dev/null @@ -XXX,XX +XXX,XX @@ -QEMU U2F Key Device Documentation. - -Contents -1. USB U2F key device -2. Building -3. Using u2f-emulated -4. Using u2f-passthru -5. Libu2f-emu - -1. USB U2F key device - -U2F is an open authentication standard that enables relying parties -exposed to the internet to offer a strong second factor option for end -user authentication. - -The standard brings many advantages to both parties, client and server, -allowing to reduce over-reliance on passwords, it increases authentication -security and simplifies passwords. - -The second factor is materialized by a device implementing the U2F -protocol. In case of a USB U2F security key, it is a USB HID device -that implements the U2F protocol. - -In QEMU, the USB U2F key device offers a dedicated support of U2F, allowing -guest USB FIDO/U2F security keys operating in two possible modes: -pass-through and emulated. - -The pass-through mode consists of passing all requests made from the guest -to the physical security key connected to the host machine and vice versa. -In addition, the dedicated pass-through allows to have a U2F security key -shared on several guests which is not possible with a simple host device -assignment pass-through. - -The emulated mode consists of completely emulating the behavior of an -U2F device through software part. Libu2f-emu is used for that. - - -2. Building - -To ensure the build of the u2f-emulated device variant which depends -on libu2f-emu: configuring and building: - - ./configure --enable-u2f && make - -The pass-through mode is built by default on Linux. To take advantage -of the autoscan option it provides, make sure you have a working libudev -installed on the host. - - -3. Using u2f-emulated - -To work, an emulated U2F device must have four elements: - * ec x509 certificate - * ec private key - * counter (four bytes value) - * 48 bytes of entropy (random bits) - -To use this type of device, this one has to be configured, and these -four elements must be passed one way or another. - -Assuming that you have a working libu2f-emu installed on the host. -There are three possible ways of configurations: - * ephemeral - * setup directory - * manual - -Ephemeral is the simplest way to configure, it lets the device generate -all the elements it needs for a single use of the lifetime of the device. - - qemu -usb -device u2f-emulated - -Setup directory allows to configure the device from a directory containing -four files: - * certificate.pem: ec x509 certificate - * private-key.pem: ec private key - * counter: counter value - * entropy: 48 bytes of entropy - - qemu -usb -device u2f-emulated,dir=$dir - -Manual allows to configure the device more finely by specifying each -of the elements necessary for the device: - * cert - * priv - * counter - * entropy - - qemu -usb -device u2f-emulated,cert=$DIR1/$FILE1,priv=$DIR2/$FILE2,counter=$DIR3/$FILE3,entropy=$DIR4/$FILE4 - - -4. Using u2f-passthru - -On the host specify the u2f-passthru device with a suitable hidraw: - - qemu -usb -device u2f-passthru,hidraw=/dev/hidraw0 - -Alternately, the u2f-passthru device can autoscan to take the first -U2F device it finds on the host (this requires a working libudev): - - qemu -usb -device u2f-passthru - - -5. Libu2f-emu - -The u2f-emulated device uses libu2f-emu for the U2F key emulation. Libu2f-emu -implements completely the U2F protocol device part for all specified -transport given by the FIDO Alliance. - -For more information about libu2f-emu see this page: -https://github.com/MattGorko/libu2f-emu. -- 2.34.1