Re: [Qemu-devel] [PATCH 11/23] hw/intc/arm_gicv3: Implement GICv3 distributor registers

From: Peter Maydell <peter.maydell@linaro.org>
To: Shannon Zhao <shannon.zhao@linaro.org>
Cc: qemu-arm <qemu-arm@nongnu.org>,
	QEMU Developers <qemu-devel@nongnu.org>,
	Patch Tracking <patches@linaro.org>,
	Pavel Fedin <p.fedin@samsung.com>,
	Shlomo Pongratz <shlomo.pongratz@huawei.com>,
	Shlomo Pongratz <shlomopongratz@gmail.com>,
	Christoffer Dall <christoffer.dall@linaro.org>
Subject: Re: [Qemu-devel] [PATCH 11/23] hw/intc/arm_gicv3: Implement GICv3 distributor registers
Date: Mon, 16 May 2016 09:56:12 +0100	[thread overview]
Message-ID: <CAFEAcA-dJA2iz=u-Omc4pYUO1TdvRsCY08-+ma1+ForF6iof_A@mail.gmail.com> (raw)
In-Reply-To: <CAFEAcA_9ZuAb79S1-PCU2kJ=6cHTYbuHTmD2VaZyecgBG_oFuw@mail.gmail.com>

On 13 May 2016 at 16:24, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 13 May 2016 at 16:05, Shannon Zhao <shannon.zhao@linaro.org> wrote:
>> So I think it shouldn't call gicv3_update if attrs.secure is true and
>> irq < 32. And it should check the parameter irq in gicv3_update().
>
> If irq < 32 then gicd_write_ipriority() will return without
> doing anything. We'll unnecessarily call gicv3_update(), but that
> does no harm, and I don't think being slightly inefficient for
> an access a correctly functioning guest will never make is a big problem.

Thinking about this again over the weekend I realised you're right,
because gicv3_update() doesn't handle out-of-range irq numbers.
I need to either make gicv3_update() return doing nothing for
out of range IRQs or ensure we don't call it in that case (and
add an assert just in case).

thanks
-- PMM