Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <ec0ce90c-b165-d84f-340d-4973b65609b3@linux.intel.com>
Date:   Fri, 2 Dec 2022 14:22:39 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Firefox/102.0 Thunderbird/102.4.2
To:     Arnd Bergmann <arnd@arndb.de>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Daehwan Jung <dh10.jung@samsung.com>
Cc:     Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Alim Akhtar <alim.akhtar@samsung.com>,
        Mathias Nyman <mathias.nyman@intel.com>,
        Linus Walleij <linus.walleij@linaro.org>,
        Colin Ian King <colin.i.king@gmail.com>,
        Artur Bujdoso <artur.bujdoso@gmail.com>,
        Juergen Gross <jgross@suse.com>,
        Tomer Maimon <tmaimon77@gmail.com>,
        "open list:USB SUBSYSTEM" <linux-usb@vger.kernel.org>,
        "open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS" 
        <devicetree@vger.kernel.org>,
        "moderated list:ARM/SAMSUNG S3C, S5P AND EXYNOS ARM ARCHITECTURES" 
        <linux-arm-kernel@lists.infradead.org>,
        "open list:ARM/SAMSUNG S3C, S5P AND EXYNOS ARM ARCHITECTURES" 
        <linux-samsung-soc@vger.kernel.org>,
        open list <linux-kernel@vger.kernel.org>, sc.suh@samsung.com,
        taehyun.cho@samsung.com, jh0801.jung@samsung.com,
        eomji.oh@samsung.com
References: <1669860811-171746-1-git-send-email-dh10.jung@samsung.com>
 <CGME20221201021942epcas2p2429ed37e1f6146b6e1a5bef23141b3f7@epcas2p2.samsung.com>
 <1669860811-171746-3-git-send-email-dh10.jung@samsung.com>
 <Y4hgnxGMEuizJumr@kroah.com>
 <c524cba6-4438-461a-ab05-9325fe09f832@app.fastmail.com>
Content-Language: en-US
From:   Mathias Nyman <mathias.nyman@linux.intel.com>
Subject: Re: [RFC PATCH v1 2/2] usb: host: add xhci-exynos to support Exynos
 SOCs
In-Reply-To: <c524cba6-4438-461a-ab05-9325fe09f832@app.fastmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 1.12.2022 11.01, Arnd Bergmann wrote:
> On Thu, Dec 1, 2022, at 09:06, Greg Kroah-Hartman wrote:
>> On Thu, Dec 01, 2022 at 11:13:31AM +0900, Daehwan Jung wrote:
>>> This driver works with xhci platform driver. It needs to override
>>> functions of xhci_plat_hc_driver. Wakelocks are used for sleep/wakeup
>>> scenario of system.
>>
>> So this means that no other platform xhci driver can be supported in the
>> same system at the same time.
>>
>> Which kind of makes sense as that's not anything a normal system would
>> have, BUT it feels very odd.  This whole idea of "override the platform
>> driver" feels fragile, why not make these just real platform drivers and
>> have the xhci platform code be a library that the other ones can use?
>> That way you have more control overall, right?

Agree that overriding the generic platform driver xhci_hc_platform_driver
from this exynos driver is odd.

But I don't understand how this works.
Where are the hcds created and added when this xhci-exonys driver binds to
the device? all this driver does in probe is the overriding?

Am I missing something here?

> 
> Agreed, having another layer here (hcd -> xhci -> xhcd_platform ->
> xhcd_exynos) would fit perfectly well into how other SoC specific
> drivers are abstracted. This could potentially also help reduce
> the amount of code duplication between other soc specific variants
> (mtk, tegra, mvebu, ...) that are all platform drivers but don't
> share code with xhci-plat.c.
> 
> Alternatively, it seems that all of the xhci-exynos support could
> just be part of the generic xhci-platform driver: as far as I can
> tell, none of the added code is exynos specific at all, instead it
> is a generic xhci that is using the wakeup_source framework.

Sounds reasonable as well, and if some exynos specific code is needed
then just create a xhci_plat_priv struct for exynos and pass it in
of_device_id data like other vendors that use the generic
xhci-platform driver do.

-Mathias