Re: [PATCH v8] i2c: virtio: add a virtio i2c frontend driver

From: Jie Deng <jie.deng@intel.com>
To: Viresh Kumar <viresh.kumar@linaro.org>, Arnd Bergmann <arnd@arndb.de>
Cc: "Michael S. Tsirkin" <mst@redhat.com>,
	"Bjorn Andersson" <bjorn.andersson@linaro.org>,
	"Wolfram Sang" <wsa+renesas@sang-engineering.com>,
	"Linux I2C" <linux-i2c@vger.kernel.org>,
	"Wolfram Sang" <wsa@kernel.org>,
	"Andy Shevchenko" <andriy.shevchenko@linux.intel.com>,
	yu1.wang@intel.com,
	"Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>,
	kblaiech@mellanox.com, virtualization@lists.linux-foundation.org,
	"Enrico Weigelt, metux IT consult" <lkml@metux.net>,
	"Stefan Hajnoczi" <stefanha@redhat.com>,
	"Tali Perry" <tali.perry1@gmail.com>,
	conghui.chen@intel.com, loic.poulain@linaro.org,
	"Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>,
	"Sergey Semin" <Sergey.Semin@baikalelectronics.ru>,
	jarkko.nikula@linux.intel.com, shuo.a.liu@intel.com,
	"Paolo Bonzini" <pbonzini@redhat.com>,
	"Mike Rapoport" <rppt@kernel.org>
Subject: Re: [PATCH v8] i2c: virtio: add a virtio i2c frontend driver
Date: Fri, 19 Mar 2021 13:31:34 +0800	[thread overview]
Message-ID: <b135b474-b167-67ad-588c-b0cfe8dc2998@intel.com> (raw)
In-Reply-To: <20210319035435.a4reve77hnvjdzwk@vireshk-i7>

On 2021/3/19 11:54, Viresh Kumar wrote:
> On 18-03-21, 15:52, Arnd Bergmann wrote:
>> Allowing multiple virtio-i2c controllers in one system, and multiple i2c
>> devices attached to each controller is clearly something that has to work.
> Good.
>
>> I don't actually see a limitation though. Viresh, what is the problem
>> you see for having multiple controllers?
> I thought this would be a problem in that case as we are using the global
> virtio_adapter here.
>
> +       vi->adap = &virtio_adapter;
> +       i2c_set_adapdata(vi->adap, vi);
>
> Multiple calls to probe() will end up updating the same pointer inside adap.
>
> +       vi->adap->dev.parent = &vdev->dev;
>
> Same here, overwrite.
>
> +       /* Setup ACPI node for controlled devices which will be probed through ACPI */
> +       ACPI_COMPANION_SET(&vi->adap->dev, ACPI_COMPANION(pdev));
> +       vi->adap->timeout = HZ / 10;
>
> These may be fine, but still not ideal I believe.
>
> +       ret = i2c_add_adapter(vi->adap);
> i
> This should be a problem as well, we must be adding this to some sort of list,
> doing some RPM stuff, etc ?
>
> Jie, the solution is to allocate memory for adap at runtime in probe and remove
> the virtio_adapter structure completely.

If you want to support that. Then I think we don't need to change the 
following at all.

> +    .algo = &virtio_algorithm,
> +
> +        return ret;
> +
> +    vi->adap = virtio_adapter;
This is strange, why are you allocating memory for adapter twice ?
Once for virtio_adapter and once for vi->adap ? Either fill the fields
directly for v->adap here and remove virtio_adapter or make vi->adap a
pointer.

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization