Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   "bibo, mao" <bibo.mao@gmail.com>
Message-ID: <cc73aaf7-85bf-2062-6d1b-f1cbb70242f8@loongson.cn>
Date:   Tue, 13 Jun 2023 20:49:26 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.12.0
Subject: Re: [PATCH v13 03/30] LoongArch: KVM: Implement kvm hardware enable,
 disable interface
To:     Tianrui Zhao <zhaotianrui@loongson.cn>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Huacai Chen <chenhuacai@kernel.org>,
        WANG Xuerui <kernel@xen0n.name>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        loongarch@lists.linux.dev, Jens Axboe <axboe@kernel.dk>,
        Mark Brown <broonie@kernel.org>,
        Alex Deucher <alexander.deucher@amd.com>,
        Oliver Upton <oliver.upton@linux.dev>,
        Xi Ruoyao <xry111@xry111.site>, tangyouling@loongson.cn,
        linux-kernel@vger.kernel.org, kvm@vger.kernel.org
References: <20230609090832.2131037-1-zhaotianrui@loongson.cn>
 <20230609090832.2131037-4-zhaotianrui@loongson.cn>
In-Reply-To: <20230609090832.2131037-4-zhaotianrui@loongson.cn>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


在 2023/6/9 17:08, Tianrui Zhao 写道:
> Implement kvm hardware enable, disable interface, setting
> the guest config register to enable virtualization features
> when called the interface.
> 
> Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
> ---
>   arch/loongarch/kvm/main.c | 64 +++++++++++++++++++++++++++++++++++++++
>   1 file changed, 64 insertions(+)
> 
> diff --git a/arch/loongarch/kvm/main.c b/arch/loongarch/kvm/main.c
> index f98c1619725f..5ebae1ea7565 100644
> --- a/arch/loongarch/kvm/main.c
> +++ b/arch/loongarch/kvm/main.c
> @@ -195,6 +195,70 @@ static void _kvm_init_gcsr_flag(void)
>   	set_gcsr_sw_flag(LOONGARCH_CSR_PERFCNTR3);
>   }
>   
> +void kvm_init_vmcs(struct kvm *kvm)
> +{
> +	kvm->arch.vmcs = vmcs;
> +}
> +
> +long kvm_arch_dev_ioctl(struct file *filp,
> +			unsigned int ioctl, unsigned long arg)
> +{
> +	return -ENOIOCTLCMD;
> +}
> +
> +#ifdef CONFIG_KVM_GENERIC_HARDWARE_ENABLING
Can we remove macro CONFIG_KVM_GENERIC_HARDWARE_ENABLING here since it 
is always enabled? The other looks good to me.

Reviewed-by: Bibo Mao <maobibo@loongson.cn>

Regards
Bibo, Mao
> +int kvm_arch_hardware_enable(void)
> +{
> +	unsigned long env, gcfg = 0;
> +
> +	env = read_csr_gcfg();
> +	/* First init gtlbc, gcfg, gstat, gintc. All guest use the same config */
> +	clear_csr_gtlbc(CSR_GTLBC_USETGID | CSR_GTLBC_TOTI);
> +	write_csr_gcfg(0);
> +	write_csr_gstat(0);
> +	write_csr_gintc(0);
> +
> +	/*
> +	 * Enable virtualization features granting guest direct control of
> +	 * certain features:
> +	 * GCI=2:       Trap on init or unimplement cache instruction.
> +	 * TORU=0:      Trap on Root Unimplement.
> +	 * CACTRL=1:    Root control cache.
> +	 * TOP=0:       Trap on Previlege.
> +	 * TOE=0:       Trap on Exception.
> +	 * TIT=0:       Trap on Timer.
> +	 */
> +	if (env & CSR_GCFG_GCIP_ALL)
> +		gcfg |= CSR_GCFG_GCI_SECURE;
> +	if (env & CSR_GCFG_MATC_ROOT)
> +		gcfg |= CSR_GCFG_MATC_ROOT;
> +
> +	gcfg |= CSR_GCFG_TIT;
> +	write_csr_gcfg(gcfg);
> +
> +	kvm_flush_tlb_all();
> +
> +	/* Enable using TGID  */
> +	set_csr_gtlbc(CSR_GTLBC_USETGID);
> +	kvm_debug("gtlbc:%llx gintc:%llx gstat:%llx gcfg:%llx",
> +			read_csr_gtlbc(), read_csr_gintc(),
> +			read_csr_gstat(), read_csr_gcfg());
> +
> +	return 0;
> +}
> +
> +void kvm_arch_hardware_disable(void)
> +{
> +	clear_csr_gtlbc(CSR_GTLBC_USETGID | CSR_GTLBC_TOTI);
> +	write_csr_gcfg(0);
> +	write_csr_gstat(0);
> +	write_csr_gintc(0);
> +
> +	/* Flush any remaining guest TLB entries */
> +	kvm_flush_tlb_all();
> +}
> +#endif
> +
>   static int kvm_loongarch_env_init(void)
>   {
>   	struct kvm_context *context;