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;
MIME-Version: 1.0
References: <20220115193245.3777833-1-laurent@vivier.eu> <20220115193245.3777833-4-laurent@vivier.eu>
 <CAK8P3a13g_o4mTsOO-4b=WU6TGRHubY7HCt1x1FdXpVmjy6-Sg@mail.gmail.com> <77e747de-c7d6-314d-e0b0-58852d463326@vivier.eu>
In-Reply-To: <77e747de-c7d6-314d-e0b0-58852d463326@vivier.eu>
From:   Arnd Bergmann <arnd@arndb.de>
Date:   Sun, 16 Jan 2022 17:19:43 +0100
Message-ID: <CAK8P3a1i_ngEaqhsy4xNrQV8=d1jY=JTxGNNUqOZ0g0to5bw8Q@mail.gmail.com>
Subject: Re: [PATCH v8 3/4] clocksource/drivers: Add a goldfish-timer clocksource
To:     Laurent Vivier <laurent@vivier.eu>
Cc:     Arnd Bergmann <arnd@arndb.de>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        John Stultz <john.stultz@linaro.org>,
        linux-rtc@vger.kernel.org, Jiaxun Yang <jiaxun.yang@flygoat.com>,
        Daniel Lezcano <daniel.lezcano@linaro.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Alessandro Zummo <a.zummo@towertech.it>,
        Alexandre Belloni <alexandre.belloni@bootlin.com>,
        Stephen Boyd <sboyd@kernel.org>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        linux-m68k <linux-m68k@lists.linux-m68k.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sun, Jan 16, 2022 at 12:54 PM Laurent Vivier <laurent@vivier.eu> wrote:
>
> Virtio has exactly the same problem in the past (prio to v1.0), and it uses __LITTLE_ENDIAN in
> virtio_legacy_is_little_endian() to know the endianness of the device.

Right, that was my point, let's not duplicate the bug from pre-standard virtio.

> So the code can become:
>
> #ifdef __LITTLE_ENDIAN
> #define goldfish_ioread32 ioread32
> #define goldfish_iowrite32 iowrite32
> #else
> #define goldfish_ioread32 ioread32be
> #define goldfish_iowrite32 iowrite32be
> #endif

This is exactly the same as what you had before, it continues to break when
host and guest disagree on the endianess.

> What do you prefer: CONFIG_M68K or __LITTLE_ENDIAN?

Checking for CONFIG_M68K at least has a chance of working, provided
that all existing qemu implementations agree on the semantics here.

       Arnd