Writes to the physical SVCR.SM and SVCR.ZA change the state of PSTATE.SM and PSTATE.ZA, causing other floating point state to reset. Emulate this behaviour for writes done via the KVM userspace ABI.
Setting PSTATE.ZA to 1 causes ZA and ZT0 to be reset to 0, these are stored in sme_state. Setting PSTATE.ZA to 0 causes ZA and ZT0 to become inaccessible so no reset is needed. Any change in PSTATE.SM causes the V, Z, P, FFR and FPMR registers to be reset to 0 and FPSR to be reset to 0x800009f. Rather than introduce a requirement that the vector configuration be finalised before writing to SVCR we check for this before updating the SVE and SME specific state, when finalisation happens they will be allocated with an initial state of 0. Similarly in order to avoid ordering requirements between finalisation and writes to the ID registers we always allocate space for ZT0 if the hardware supports it, this is 512 bytes per vCPU. The overwhelming majority of practical systems with SME are expected to want use SME2, there is very little practical reason to disable it other than for feature testing, and the additional complexity seems more likely to lead to bugs than deliver practical benefits. Signed-off-by: Mark Brown <[email protected]> --- arch/arm64/include/asm/kvm_host.h | 28 ++++++++++++++++++++++++++++ arch/arm64/include/asm/sysreg.h | 2 ++ arch/arm64/kvm/sys_regs.c | 30 +++++++++++++++++++++++++++++- 3 files changed, 59 insertions(+), 1 deletion(-) diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index 35339cbf23f9..b78c039cb5ec 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -1142,6 +1142,34 @@ struct kvm_vcpu_arch { #define vcpu_sve_state_size(vcpu) sve_state_size_from_vl(vcpu_sve_max_vl(vcpu)) +#define vcpu_sme_state(vcpu) (kern_hyp_va((vcpu)->arch.sme_state)) + +#define sme_state_size_from_vl(vl, sme2) ({ \ + size_t __size_ret; \ + unsigned int __vq; \ + \ + if (WARN_ON(!sve_vl_valid(vl))) { \ + __size_ret = 0; \ + } else { \ + __vq = sve_vq_from_vl(vl); \ + __size_ret = ZA_SIG_REGS_SIZE(__vq); \ + if (sme2) \ + __size_ret += ZT_SIG_REG_BYTES; \ + } \ + \ + __size_ret; \ +}) + +/* + * Always provide space for ZT0 to avoid ordering requirements with ID + * register writes and vector finalization. + */ +#define vcpu_sme_state_size(vcpu) ({ \ + unsigned long __vl; \ + __vl = (vcpu)->arch.max_vl[ARM64_VEC_SME]; \ + sme_state_size_from_vl(__vl, system_supports_sme2()); \ +}) + #define KVM_GUESTDBG_VALID_MASK (KVM_GUESTDBG_ENABLE | \ KVM_GUESTDBG_USE_SW_BP | \ KVM_GUESTDBG_USE_HW | \ diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h index 4b96449e0ffa..b434320de1a7 100644 --- a/arch/arm64/include/asm/sysreg.h +++ b/arch/arm64/include/asm/sysreg.h @@ -1108,6 +1108,8 @@ #define gicr_insn(insn) read_sysreg_s(GICV5_OP_GICR_##insn) #define gic_insn(v, insn) write_sysreg_s(v, GICV5_OP_GIC_##insn) +#define FPSR_RESET_VALUE 0x800009f + #ifdef __ASSEMBLER__ .macro mrs_s, rt, sreg diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c index c43cb1b8fb68..e8d3eceb0124 100644 --- a/arch/arm64/kvm/sys_regs.c +++ b/arch/arm64/kvm/sys_regs.c @@ -1015,6 +1015,34 @@ static unsigned int hidden_visibility(const struct kvm_vcpu *vcpu, return REG_HIDDEN; } +static int set_svcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd, + u64 val) +{ + u64 old = __vcpu_sys_reg(vcpu, rd->reg); + + if (val & SVCR_RES0) + return -EINVAL; + + if ((val & SVCR_ZA) && !(old & SVCR_ZA) && + kvm_arm_vcpu_vec_finalized(vcpu)) + memset(vcpu->arch.sme_state, 0, vcpu_sme_state_size(vcpu)); + + if ((val & SVCR_SM) != (old & SVCR_SM)) { + memset(vcpu->arch.ctxt.fp_regs.vregs, 0, + sizeof(vcpu->arch.ctxt.fp_regs.vregs)); + + if (kvm_arm_vcpu_vec_finalized(vcpu)) + memset(vcpu->arch.sve_state, 0, + vcpu_sve_state_size(vcpu)); + + __vcpu_assign_sys_reg(vcpu, FPMR, 0); + vcpu->arch.ctxt.fp_regs.fpsr = FPSR_RESET_VALUE; + } + + __vcpu_assign_sys_reg(vcpu, rd->reg, val); + return 0; +} + static unsigned int pmu_visibility(const struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) { @@ -3612,7 +3640,7 @@ static const struct sys_reg_desc sys_reg_descs[] = { CTR_EL0_DminLine_MASK | CTR_EL0_L1Ip_MASK | CTR_EL0_IminLine_MASK), - { SYS_DESC(SYS_SVCR), undef_access, reset_val, SVCR, 0, .visibility = sme_visibility }, + { SYS_DESC(SYS_SVCR), undef_access, reset_val, SVCR, 0, .visibility = sme_visibility, .set_user = set_svcr }, { SYS_DESC(SYS_FPMR), undef_access, reset_val, FPMR, 0, .visibility = fp8_visibility }, { PMU_SYS_REG(PMCR_EL0), .access = access_pmcr, .reset = reset_pmcr, -- 2.47.3
