linux-kernel - Re: [PATCH v2 01/29] LoongArch: KVM: Add kvm related header files

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 20 Feb 2023 19:22:11 +0100
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Tianrui Zhao <zhaotianrui@...ngson.cn>
Cc:     Huacai Chen <chenhuacai@...nel.org>,
        WANG Xuerui <kernel@...0n.name>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        loongarch@...ts.linux.dev, linux-kernel@...r.kernel.org,
        kvm@...r.kernel.org, Jens Axboe <axboe@...nel.dk>,
        Mark Brown <broonie@...nel.org>,
        Alex Deucher <alexander.deucher@....com>,
        Oliver Upton <oliver.upton@...ux.dev>, maobibo@...ngson.cn
Subject: Re: [PATCH v2 01/29] LoongArch: KVM: Add kvm related header files

On 2/20/23 07:57, Tianrui Zhao wrote:
> +
> +/* Resume Flags */
> +#define RESUME_FLAG_DR		(1<<0)	/* Reload guest nonvolatile state? */
> +#define RESUME_FLAG_HOST	(1<<1)	/* Resume host? */
> +
> +#define RESUME_GUEST		0
> +#define RESUME_GUEST_DR		RESUME_FLAG_DR
> +#define RESUME_HOST		RESUME_FLAG_HOST
> +

Most of this code is dead, I'll give more instructions in a reply to 
patch 8.

> +	unsigned long guest_eentry;
> +	unsigned long host_eentry;
> +	int (*vcpu_run)(struct kvm_run *run, struct kvm_vcpu *vcpu);
> +	int (*handle_exit)(struct kvm_run *run, struct kvm_vcpu *vcpu);
> +
> +	/* Host registers preserved across guest mode execution */
> +	unsigned long host_stack;
> +	unsigned long host_gp;
> +	unsigned long host_pgd;
> +	unsigned long host_pgdhi;
> +	unsigned long host_entryhi;
> +
> +	/* Host CSR registers used when handling exits from guest */
> +	unsigned long badv;
> +	unsigned long host_estat;
> +	unsigned long badi;
> +	unsigned long host_ecfg;
> +	unsigned long host_percpu;
> +
> +	/* GPRS */
> +	unsigned long gprs[32];
> +	unsigned long pc;
> +
> +	/* FPU State */
> +	struct loongarch_fpu fpu FPU_ALIGN;
> +	/* Which auxiliary state is loaded (KVM_LOONGARCH_AUX_*) */
> +	unsigned int aux_inuse;
> +
> +	/* CSR State */
> +	struct loongarch_csrs *csr;
> +
> +	/* GPR used as IO source/target */
> +	u32 io_gpr;
> +
> +	struct hrtimer swtimer;
> +	/* Count timer control KVM register */
> +	u32 count_ctl;
> +
> +	/* Bitmask of exceptions that are pending */
> +	unsigned long irq_pending;
> +	/* Bitmask of pending exceptions to be cleared */
> +	unsigned long irq_clear;
> +
> +	/* Cache some mmu pages needed inside spinlock regions */
> +	struct kvm_mmu_memory_cache mmu_page_cache;
> +
> +	/* vcpu's vpid is different on each host cpu in an smp system */
> +	u64 vpid[NR_CPUS];

In _kvm_check_vmid(), you already have

+	if (migrated || (ver != old)) {
+		_kvm_update_vpid(vcpu, cpu);
+		trace_kvm_vpid_change(vcpu, vcpu->arch.vpid[cpu]);
+	}

so a vpid will never be recycled if a vCPU migrates from physical CPU A 
to B and back to A.

So please keep the current VPID in the per-cpu struct vmcs, and you can 
just copy it from there in _kvm_check_vmid().

> +	/* Period of stable timer tick in ns */
> +	u64 timer_period;
> +	/* Frequency of stable timer in Hz */
> +	u64 timer_mhz;
> +	/* Stable bias from the raw time */
> +	u64 timer_bias;
> +	/* Dynamic nanosecond bias (multiple of timer_period) to avoid overflow */
> +	s64 timer_dyn_bias;
> +	/* Save ktime */
> +	ktime_t stable_ktime_saved;
> +
> +	u64 core_ext_ioisr[4];
> +
> +	/* Last CPU the VCPU state was loaded on */
> +	int last_sched_cpu;
> +	/* Last CPU the VCPU actually executed guest code on */
> +	int last_exec_cpu;
> +
> +	u8 fpu_enabled;

This field is always true, please remove it.

> +	struct kvm_guest_debug_arch guest_debug;

This struct is empty, please remove it.

Paolo