Re: [Qemu-devel] [Qemu-arm] [PATCH 3/3] hw/misc/exynos4210_pmu: Reorder local variables for readability

From: Peter Maydell <peter.maydell@linaro.org>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Philippe Mathieu-Daudé" <f4bug@amsat.org>,
	"Krzysztof Kozlowski" <krzk@kernel.org>,
	"Igor Mitsyanko" <i.mitsyanko@gmail.com>,
	qemu-arm <qemu-arm@nongnu.org>,
	"QEMU Developers" <qemu-devel@nongnu.org>
Subject: Re: [Qemu-devel] [Qemu-arm] [PATCH 3/3] hw/misc/exynos4210_pmu: Reorder local variables for readability
Date: Mon, 6 Mar 2017 13:16:15 +0000	[thread overview]
Message-ID: <CAFEAcA-6yQ=95hGdia6TDhq4C-swOJ1iZA8C+u99Dy9kfG+f0w@mail.gmail.com> (raw)
In-Reply-To: <4989d5b6-586a-de18-e8e6-6ab9e9ac131c@redhat.com>

On 6 March 2017 at 13:03, Paolo Bonzini <pbonzini@redhat.com> wrote:
>
>
> On 06/03/2017 10:46, Peter Maydell wrote:
>> On 6 March 2017 at 04:06, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>> your change seems OK but while you are here, 'unsigned' is considered
>>> harmful since more than a decade.
>>
>> Considered harmful by who, and why?
>
> It is true that unsigned has all the disadvantages of "int" (it may be
> smaller than the size of the object" and all the disadvantages of
> "size_t" (it doesn't optimize as well because the compiler cannot
> exploit undefined behavior).
>
> I wouldn't call it harmful, but it feels like the worst of both worlds.

size_t is a pretty silly choice for a variable that's looping
through the number of registers in the device, which is a value
that would fit in 16 bits, let alone 32. I would probably have
written this with 'int', but use of 'unsigned' doesn't come
very high up on my list of things to complain about.

(since we use -fwrapv the compiler can't exploit undefined
behaviour for signed loop indexes either.)

thanks
-- PMM