Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 11 Nov 2020 13:45:27 +0000
From:   David Brazdil <dbrazdil@google.com>
To:     Marc Zyngier <maz@kernel.org>
Cc:     kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org,
        linux-kernel@vger.kernel.org, James Morse <james.morse@arm.com>,
        Julien Thierry <julien.thierry.kdev@gmail.com>,
        Suzuki K Poulose <suzuki.poulose@arm.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will@kernel.org>, Dennis Zhou <dennis@kernel.org>,
        Tejun Heo <tj@kernel.org>, Christoph Lameter <cl@linux.com>,
        Mark Rutland <mark.rutland@arm.com>,
        Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
        Quentin Perret <qperret@google.com>,
        Andrew Scull <ascull@google.com>,
        Andrew Walbran <qwandor@google.com>, kernel-team@android.com
Subject: Re: [PATCH v1 07/24] kvm: arm64: Create nVHE copy of cpu_logical_map
Message-ID: <20201111134527.ygzf5cm6qpdpcroo@google.com>
References: <20201109113233.9012-1-dbrazdil@google.com>
 <20201109113233.9012-8-dbrazdil@google.com>
 <d473fd26e5314f2407b70242488f33de@kernel.org>
 <20201111130321.qalrzfabdonrwvsz@google.com>
 <15a580e6ac06294ead8859fba8f51deb@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <15a580e6ac06294ead8859fba8f51deb@kernel.org>
Precedence: bulk

On Wed, Nov 11, 2020 at 01:29:29PM +0000, Marc Zyngier wrote:
> On 2020-11-11 13:03, David Brazdil wrote:
> > > > +/*
> > > > + * nVHE copy of data structures tracking available CPU cores.
> > > > + * Only entries for CPUs that were online at KVM init are populated.
> > > > + * Other CPUs should not be allowed to boot because their features were
> > > > + * not checked against the finalized system capabilities.
> > > > + */
> > > > +u64 __ro_after_init __cpu_logical_map[NR_CPUS] = { [0 ... NR_CPUS-1]
> > > > = INVALID_HWID };
> > > 
> > > I'm not sure what __ro_after_init means once we get S2 isolation.
> > 
> > It is stretching the definition of 'init' a bit, I know, but I don't see
> > what
> > your worry is about S2? The intention is to mark this read-only for
> > .hyp.text
> > at runtime. With S2, the host won't be able to write to it after KVM
> > init.
> > Obviously that's currently not the case.
> 
> More importantly, EL2 can write to it at any time, which is the bit I'm
> worried
> about, as it makes the annotation misleading.

EL2 can't, at least not accidentally. The hyp memory mapping is PAGE_HYP_RO
(see patch 05). Does this annotation have stronger guarantees in EL1?
AFAICT, those variables are made PAGE_KERNEL_RO in mark_rodata_ro().

> 
> > One thing we might change in the future is marking it RW for some
> > initial
> > setup in a HVC handler, then marking it RO for the rest of uptime.
> 
> That'd be a desirable outcome, and it would be consistent with the rest
> of the kernel.
> 
> > 
> > > 
> > > > +
> > > > +u64 cpu_logical_map(int cpu)
> > > 
> > > nit: is there any reason why "cpu" cannot be unsigned? The thought
> > > of a negative CPU number makes me shiver...
> > 
> > Same here. That's how it's defined in kernel proper, so I went with
> > that.
> 
> I'm happy to deviate from the kernel (give the function a different name
> if this clashes with existing include files).
> 
> We can also fix the rest of the kernel (I've just written the trivial
> patch).

Shouldn't clash with include files. Where fixing the kernel might clash is
all the users of for_each_*_cpu that use an int for the iterator var.