| Message ID | 20230530194623.272652-13-dbarboza@ventanamicro.com |
|---|---|
| State | New |
| Headers | show |
| Series | target/riscv, KVM: fixes and enhancements | expand |
On Tue, May 30, 2023 at 04:46:19PM -0300, Daniel Henrique Barboza wrote: > Our design philosophy with KVM properties can be resumed in two main > decisions based on KVM interface availability and what the user wants to > do: > > - if the user disables an extension that the host KVM module doesn't > know about (i.e. it doesn't implement the kvm_get_one_reg() interface), > keep booting the CPU. This will avoid users having to deal with issues > with older KVM versions while disabling features they don't care; > > - for any other case we're going to error out immediately. If the user > wants to enable a feature that KVM doesn't know about this a problem that > is worth aborting - the user must know that the feature wasn't enabled > in the hart. Likewise, if KVM knows about the extension, the user wants > to enable/disable it, and we fail to do it so, that's also a problem we > can't shrug it off. > > For MISA bits we're going to be a little more conservative: we won't > even try enabling bits that aren't already available in the host. The > ioctl() is so likely to fail that's not worth trying. This check is > already done in the previous patch, in kvm_cpu_set_misa_ext_cfg(), thus > we don't need to worry about it now. > > In kvm_riscv_update_cpu_misa_ext() we'll go through every potential user > option and do as follows: > > - if the user didn't set the property or set to the same value of the > host, do nothing; > > - Disable the given extension in KVM. If it fails we need to verify the > error code. -EINVAL indicates that KVM doesn't know about the reg, so > re-enable the extension in env->misa_ext and keep booting. If it fails We shouldn't "re-enable the extension in env->misa_ext..." when the extension isn't supported by KVM for any reason. But, assuming EINVAL is only returned when KVM doesn't support the extension (I wish it returned ENOENT instead), then we'll never get EINVAL here in update anyway. env->misa_ext is initialized to what KVM supports, so it wouldn't have had this unsupported extension bit set in the first place, meaning 'user_set' wouldn't get set at property setting time either. > for any other reason we're going to exit out. > > Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com> > --- > target/riscv/kvm.c | 41 +++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 41 insertions(+) > > diff --git a/target/riscv/kvm.c b/target/riscv/kvm.c > index 6afd56cda5..bb1dafe263 100644 > --- a/target/riscv/kvm.c > +++ b/target/riscv/kvm.c > @@ -166,6 +166,42 @@ static void kvm_cpu_set_misa_ext_cfg(Object *obj, Visitor *v, > "enabled in the host", misa_ext_cfg->name); > } > > +static void kvm_riscv_update_cpu_misa_ext(RISCVCPU *cpu, CPUState *cs) > +{ > + CPURISCVState *env = &cpu->env; > + uint64_t id, reg; > + int i, ret; > + > + for (i = 0; i < ARRAY_SIZE(kvm_misa_ext_cfgs); i++) { > + RISCVCPUMisaExtConfig *misa_cfg = &kvm_misa_ext_cfgs[i]; > + > + if (!misa_cfg->user_set) { > + continue; > + } > + > + /* If we're here we're going to disable the MISA bit */ > + reg = 0; > + id = kvm_riscv_reg_id(env, KVM_REG_RISCV_ISA_EXT, > + misa_cfg->kvm_reg_id); > + ret = kvm_set_one_reg(cs, id, ®); > + if (ret != 0) { > + if (ret == -EINVAL) { > + /* > + * KVM doesn't know how to handle this bit. Since > + * it's an extension that the user wants to disable, > + * do not error out. > + */ > + continue; This case can be replaced with a comment explaining we don't ever expect EINVAL here at update time, since user_set will never be true for user-disabled extensions which KVM doesn't support. > + } else { > + error_report("Unable to set KVM reg %s, error %d", > + misa_cfg->name, ret); > + exit(EXIT_FAILURE); > + } > + } > + env->misa_ext &= ~misa_cfg->misa_bit; > + } > +} > + > static void kvm_riscv_add_cpu_user_properties(Object *cpu_obj) > { > int i; > @@ -632,8 +668,13 @@ int kvm_arch_init_vcpu(CPUState *cs) > > if (!object_dynamic_cast(OBJECT(cpu), TYPE_RISCV_CPU_HOST)) { > ret = kvm_vcpu_set_machine_ids(cpu, cs); > + if (ret != 0) { > + return ret; > + } > } > > + kvm_riscv_update_cpu_misa_ext(cpu, cs); > + > return ret; > } > > -- > 2.40.1 > > Thanks, drew
diff --git a/target/riscv/kvm.c b/target/riscv/kvm.c index 6afd56cda5..bb1dafe263 100644 --- a/target/riscv/kvm.c +++ b/target/riscv/kvm.c @@ -166,6 +166,42 @@ static void kvm_cpu_set_misa_ext_cfg(Object *obj, Visitor *v, "enabled in the host", misa_ext_cfg->name); } +static void kvm_riscv_update_cpu_misa_ext(RISCVCPU *cpu, CPUState *cs) +{ + CPURISCVState *env = &cpu->env; + uint64_t id, reg; + int i, ret; + + for (i = 0; i < ARRAY_SIZE(kvm_misa_ext_cfgs); i++) { + RISCVCPUMisaExtConfig *misa_cfg = &kvm_misa_ext_cfgs[i]; + + if (!misa_cfg->user_set) { + continue; + } + + /* If we're here we're going to disable the MISA bit */ + reg = 0; + id = kvm_riscv_reg_id(env, KVM_REG_RISCV_ISA_EXT, + misa_cfg->kvm_reg_id); + ret = kvm_set_one_reg(cs, id, ®); + if (ret != 0) { + if (ret == -EINVAL) { + /* + * KVM doesn't know how to handle this bit. Since + * it's an extension that the user wants to disable, + * do not error out. + */ + continue; + } else { + error_report("Unable to set KVM reg %s, error %d", + misa_cfg->name, ret); + exit(EXIT_FAILURE); + } + } + env->misa_ext &= ~misa_cfg->misa_bit; + } +} + static void kvm_riscv_add_cpu_user_properties(Object *cpu_obj) { int i; @@ -632,8 +668,13 @@ int kvm_arch_init_vcpu(CPUState *cs) if (!object_dynamic_cast(OBJECT(cpu), TYPE_RISCV_CPU_HOST)) { ret = kvm_vcpu_set_machine_ids(cpu, cs); + if (ret != 0) { + return ret; + } } + kvm_riscv_update_cpu_misa_ext(cpu, cs); + return ret; }
Our design philosophy with KVM properties can be resumed in two main decisions based on KVM interface availability and what the user wants to do: - if the user disables an extension that the host KVM module doesn't know about (i.e. it doesn't implement the kvm_get_one_reg() interface), keep booting the CPU. This will avoid users having to deal with issues with older KVM versions while disabling features they don't care; - for any other case we're going to error out immediately. If the user wants to enable a feature that KVM doesn't know about this a problem that is worth aborting - the user must know that the feature wasn't enabled in the hart. Likewise, if KVM knows about the extension, the user wants to enable/disable it, and we fail to do it so, that's also a problem we can't shrug it off. For MISA bits we're going to be a little more conservative: we won't even try enabling bits that aren't already available in the host. The ioctl() is so likely to fail that's not worth trying. This check is already done in the previous patch, in kvm_cpu_set_misa_ext_cfg(), thus we don't need to worry about it now. In kvm_riscv_update_cpu_misa_ext() we'll go through every potential user option and do as follows: - if the user didn't set the property or set to the same value of the host, do nothing; - Disable the given extension in KVM. If it fails we need to verify the error code. -EINVAL indicates that KVM doesn't know about the reg, so re-enable the extension in env->misa_ext and keep booting. If it fails for any other reason we're going to exit out. Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com> --- target/riscv/kvm.c | 41 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 41 insertions(+)