Re: [PATCH v2 3/3] virtio-spi: Add virtio SPI driver.

[Harald Mommer]

Hello,

virtio-dev still down...

I have to admit when it comes to device trees it's most of the time a 
fight. You will see below, fighting.

On 18.03.24 10:39, Haixu Cui wrote:
> Hello,
>
>     As Mark recommended, I reserve the virtio_spi_probe function only 
> and list my comments.
>
> On 3/4/2024 11:43 PM, Harald Mommer wrote:
> > +static int virtio_spi_probe(struct virtio_device *vdev)
> > +{
> > +    struct device_node *np = vdev->dev.parent->of_node;
> > +    struct virtio_spi_priv *priv;
> > +    struct spi_controller *ctrl;
> > +    int err;
> > +    u32 bus_num;
> > +    u16 csi;
> > +
> > +    ctrl = devm_spi_alloc_host(&vdev->dev, sizeof(*priv));
> > +    if (!ctrl)
> > +        return -ENOMEM;
> > +
> > +    priv = spi_controller_get_devdata(ctrl);
> > +    priv->vdev = vdev;
> > +    vdev->priv = priv;
> > +    dev_set_drvdata(&vdev->dev, ctrl);
>
>     ctrl->dev.of_node is not set, so the child node cannot be parsed. 
> I would say, it's necessary to support the virtio spi device node in 
> the format:


What you most probably want to have here is

  ctrl->dev.of_node = np;

for reasons I don't understand completely yet but have some idea. I may 
play around to get it.


>     virtio-spi@4b013000 {
>         compatible = "virtio,mmio";
>         reg = <0x4b013000 0x1000>;
>         interrupts = <0 497 4>;
>
>         spi {
>             compatible = "virtio,device2d";
>             #address-cells = <1>;
>             #size-cells = <0>;
>
>             spidev {
>                 compatible = "xxx";
>                 reg = <0>;
>                 spi-max-frequency = <100000>;
>             };
>         };
>     };


Looks in my setup like this:

virtio_mmio@4b013000 {
  compatible = "virtio,mmio";
  reg = <0x0 0x4b013000 0x0 0x1000>;
  /* GIC_SPI = 0, IRQ_TYPE_LEVEL_HIGH = 4 */
  interrupts = <0 497 4>;
  spi,bus-num = <1234>; /* <<<=== This here has been added */
};

The "spi,bus-num" is missing in your setup so you use the default of 0.

For what is "reg = <0>;" good? Or is it a dummy and only has to be present?

I don't understand spi-max-frequency in the device tree entry as we get 
this value from the config space.


> > +
> > +    init_completion(&priv->spi_req.completion);
> > +
> > +    err = of_property_read_u32(np, "spi,bus-num", &bus_num);


np already set so this assignment should play no role here when reading 
the entry.


> > +    if (!err && bus_num <= S16_MAX)
> > +        ctrl->bus_num = (s16)bus_num;
> > +
> > +    virtio_spi_read_config(vdev);
> > +
> > +    ctrl->transfer_one = virtio_spi_transfer_one;
> > +
> > +    err = virtio_spi_find_vqs(priv);
> > +    if (err) {
> > +        dev_err(&vdev->dev, "Cannot setup virtqueues\n");
> > +        return err;
> > +    }
> > +
> > +    err = spi_register_controller(ctrl);
> > +    if (err) {
> > +        dev_err(&vdev->dev, "Cannot register controller\n");
> > +        goto err_return;
> > +    }
> > +
> > +    board_info.max_speed_hz = priv->max_freq_hz;
> > +    /* spi_new_device() currently does not use bus_num but better 
> > set it */
> > +    board_info.bus_num = (u16)ctrl->bus_num;
> > +
> > +    /* Add chip selects to controller */
>
>         Creating the spi devices here statically, will introduce 
> conflicts if the same spi devices also in the device-tree.


Of course I need to create the SPI devices here statically. The 
"spi,bus-num = <1234>;" in the device tree has to be defined in a way 
that there are no conflicts. I do not understand this.


> > +    for (csi = 0; csi < ctrl->num_chipselect; csi++) {
> > +        dev_dbg(&vdev->dev, "Setting up CS=%u\n", csi);
> > +        board_info.chip_select = csi;
> > +
> > +        if (!(priv->mode_func_supported & VIRTIO_SPI_CS_HIGH))
> > +            board_info.mode = SPI_MODE_0;
> > +        else
> > +            board_info.mode = SPI_MODE_0 | SPI_CS_HIGH;
> > +
> > +        if (!spi_new_device(ctrl, &board_info)) {
> > +            dev_err(&vdev->dev, "Cannot setup device %u\n", csi);
> > +            spi_unregister_controller(ctrl);
> > +            err = -ENODEV;
> > +            goto err_return;
> > +        }
> > +    }
> > +
> > +    return 0;
> > +
> > +err_return:
> > +    vdev->config->del_vqs(vdev);
> > +    return err;
> > +}
> > +

What I needed to support for our needs was a user space SPI interface. 
You probably have a different setup and additional more complex device 
tree configuration needs I do not understand (yet). I'm somewhat lost in 
the moment.