[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <ZkOLz0ANkESSv5qr@debug.ba.rivosinc.com>
Date: Tue, 14 May 2024 09:05:35 -0700
From: Deepak Gupta <debug@...osinc.com>
To: Clément Léger <cleger@...osinc.com>
Cc: Conor Dooley <conor@...nel.org>, Rob Herring <robh+dt@...nel.org>,
Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
Paul Walmsley <paul.walmsley@...ive.com>,
Palmer Dabbelt <palmer@...belt.com>,
Albert Ou <aou@...s.berkeley.edu>, Anup Patel <anup@...infault.org>,
Atish Patra <atishp@...shpatra.org>,
linux-riscv@...ts.infradead.org, devicetree@...r.kernel.org,
linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
kvm-riscv@...ts.infradead.org, Ved Shanbhogue <ved@...osinc.com>
Subject: Re: [RFC PATCH 6/7] riscv: kvm: add SBI FWFT support for
SBI_FWFT_DOUBLE_TRAP_ENABLE
On Tue, May 14, 2024 at 11:43:15AM +0200, Clément Léger wrote:
>
>
>On 27/04/2024 03:17, Deepak Gupta wrote:
>> On Thu, Apr 18, 2024 at 04:26:45PM +0200, Clément Léger wrote:
>>> Add support in KVM SBI FWFT extension to allow VS-mode to request double
>>> trap enabling. Double traps can then be generated by VS-mode, allowing
>>> M-mode to redirect them to S-mode.
>>>
>>
>>> +
>>> + if (value)
>>> + csr_set(CSR_HENVCFG_DBLTRP, DBLTRP_DTE);
>>> + else
>>> + csr_clear(CSR_HENVCFG_DBLTRP, DBLTRP_DTE);
>>
>> I think vcpu->arch.cfg has `henvcfg` field. Can we reflect it there as
>> well so that current
>> `henvcfg` copy in vcpu arch specifci config is consistent? Otherwise
>> it'll be lost when vCPU
>> is scheduled out and later scheduled back in (on vcpu load)
>
>henvcfg is restored when loading the vpcu (kvm_arch_vcpu_load()) and
>saved when the CPU is put (kvm_arch_vcpu_put()). But I just saw that
>this change is included in the next patch. Should have been this one ,
>I'll fix that.
>
>
>>
>> Furthermore, lets not do feature specific alias names for CSR.
>>
>> Instead let's keep consistent 64bit image of henvcfg in vcpu->arch.cfg.
>>
>> And whenever it's time to pick up the setting, pick up logic either
>> perform the writes in
>> henvcfg. And if required it'll perform henvcfgh too (as
>> `kvm_arch_vcpu_load` already does)
>
>I don't have a strong opinion on that point so if you think it really is
>better, I'll switch to that.
Thanks.
>
>Thanks,
>
>Clément
>
>>
>>> +
>>> + return SBI_SUCCESS;
>>> +}
>>> +
>>> +static int kvm_sbi_fwft_get_double_trap(struct kvm_vcpu *vcpu,
>>> + struct kvm_sbi_fwft_config *conf,
>>> + unsigned long *value)
>>> +{
>>> + if (!riscv_double_trap_enabled())
>>> + return SBI_ERR_NOT_SUPPORTED;
>>> +
>>> + *value = (csr_read(CSR_HENVCFG_DBLTRP) & DBLTRP_DTE) != 0;
>>> +
>>> + return SBI_SUCCESS;
>>> +}
>>> +
>>> static struct kvm_sbi_fwft_config *
>>> kvm_sbi_fwft_get_config(struct kvm_vcpu *vcpu, enum sbi_fwft_feature_t
>>> feature)
>>> {
>>> @@ -111,6 +147,11 @@ static const struct kvm_sbi_fwft_feature
>>> features[] = {
>>> .id = SBI_FWFT_MISALIGNED_DELEG,
>>> .set = kvm_sbi_fwft_set_misaligned_delegation,
>>> .get = kvm_sbi_fwft_get_misaligned_delegation,
>>> + },
>>> + {
>>> + .id = SBI_FWFT_DOUBLE_TRAP_ENABLE,
>>> + .set = kvm_sbi_fwft_set_double_trap,
>>> + .get = kvm_sbi_fwft_get_double_trap,
>>> }
>>> };
>>>
>>> --
>>> 2.43.0
>>>
>>>
Powered by blists - more mailing lists