Re: [PATCH v2 4/7] USB: serial: xr: add support for XR21B142X devices

From: Johan Hovold
Date: Tue Mar 30 2021 - 11:04:51 EST

Next message: Xiongwei Song: "[PATCH v2] powerpc/traps: Enhance readability for trap types"
Previous message: Miroslav Benes: "Re: [PATCH v3 00/16] x86,objtool: Optimize !RETPOLINE"
In reply to: Mauro Carvalho Chehab: "[PATCH v2 4/7] USB: serial: xr: add support for XR21B142X devices"
Next in thread: Mauro Carvalho Chehab: "[PATCH v2 1/7] USB: serial: xr: simplify its namespace"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Wed, Mar 24, 2021 at 08:41:08AM +0100, Mauro Carvalho Chehab wrote:
> Those devices can have 1, 2 or 4 ports. They're similar to
> XR21B141X models, except for:
>
> - the register numbers and reqs are different;
> - some settings are done via the control interface, by
> using CDC commands.
> - a few other minor differences.
>
> The same way as XR21B141X, the control and data interfaces
> are grouped via CDC union, where the even interface is for
> control, and the subsequent one for data. So, on XR21B142X,
>
> CDC Union:
> bMasterInterface 0
> bSlaveInterface 1
>
> bDataInterface 1
>
> CDC Union:
> bMasterInterface 0
> bSlaveInterface 1
>
> bDataInterface 1
>
> CDC Union:
> bMasterInterface 2
> bSlaveInterface 3
>
> bDataInterface 3
>
> CDC Union:
> bMasterInterface 4
> bSlaveInterface 5
>
> bDataInterface 5
>
> CDC Union:
> bMasterInterface 6
> bSlaveInterface 7
>
> bDataInterface 7
>
> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@xxxxxxxxxx>
> ---
> drivers/usb/serial/xr_serial.c | 235 ++++++++++++++++++++++++++++++++-
> 1 file changed, 231 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/usb/serial/xr_serial.c b/drivers/usb/serial/xr_serial.c
> index f161394d9b2d..058624d15505 100644
> --- a/drivers/usb/serial/xr_serial.c
> +++ b/drivers/usb/serial/xr_serial.c
> @@ -16,6 +16,7 @@
> #include <linux/slab.h>
> #include <linux/tty.h>
> #include <linux/usb.h>
> +#include <linux/usb/cdc.h>
> #include <linux/usb/serial.h>
>
> struct xr_txrx_clk_mask {
> @@ -89,8 +90,17 @@ struct xr_txrx_clk_mask {
> #define UART_MODE_RS485 0x3
> #define UART_MODE_RS485_ADDR 0x4
>
> +/* Used on devices that need the CDC control interface */
> +#define URM_RESET_RX_FIFO_BASE 0x18
> +#define URM_RESET_TX_FIFO_BASE 0x1c

The driver already has defines for these.

> +#define CDC_DATA_INTERFACE_TYPE 0x0a
> +
> +#define VIA_CDC_REGISTER -1
> +
> enum xr_model {
> XR21V141X,
> + XR21B142X,
> MAX_XR_MODELS
> };
>
> @@ -140,12 +150,37 @@ static const int xr_hal_table[MAX_XR_MODELS][MAX_XR_HAL_TYPE] = {
>
> [REQ_SET] = 0,
> [REQ_GET] = 1,
> + },
> + [XR21B142X] = {
> + [REG_ENABLE] = 0x00,
> + [REG_FORMAT] = VIA_CDC_REGISTER,

Here you're mixing register addresses in with boolean flags in an
int-array. A structure seems more appropriate.

> + [REG_FLOW_CTRL] = 0x06,
> + [REG_XON_CHAR] = 0x07,
> + [REG_XOFF_CHAR] = 0x08,
> + [REG_TX_BREAK] = 0x0a,
> + [REG_RS485_DELAY] = 0x0b,
> + [REG_GPIO_MODE] = 0x0c,
> + [REG_GPIO_DIR] = 0x0d,
> + [REG_GPIO_SET] = 0x0e,
> + [REG_GPIO_CLR] = 0x0f,
> + [REG_GPIO_STATUS] = 0x10,
> + [REG_GPIO_INT_MASK] = 0x11,
> + [REG_CUSTOMIZED_INT] = 0x12,
> + [REG_GPIO_PULL_UP_ENABLE] = 0x14,
> + [REG_GPIO_PULL_DOWN_ENABLE] = 0x15,
> + [REG_LOOPBACK] = 0x16,
> + [REG_LOW_LATENCY] = 0x46,
> + [REG_CUSTOM_DRIVER] = 0x60,
> +
> + [REQ_SET] = 0,
> + [REQ_GET] = 0,
> }
> };
>
> struct xr_port_private {
> enum xr_model model;
> unsigned int channel;
> + struct usb_interface *control_if;
> };
>
> static int xr_set_reg(struct usb_serial_port *port, u8 block, u8 reg, u8 val)
> @@ -159,6 +194,9 @@ static int xr_set_reg(struct usb_serial_port *port, u8 block, u8 reg, u8 val)
> if (port_priv->channel)
> reg |= (port_priv->channel - 1) << 8;
> break;
> + case XR21B142X:
> + reg |= (port_priv->channel - 4) << 1;

This just look really weird. This type uses the control interface number
as channel index.

But more importantly, you don't increase the register width which is
16-bit for XR21B142X and the request recipient is the interface, not the
device for this type.

> static int xr_set_reg_uart(struct usb_serial_port *port, u8 reg, u8 val)
> {
> return xr_set_reg(port, UART_REG_BLOCK, reg, val);
> @@ -243,6 +321,11 @@ static int xr_uart_enable(struct usb_serial_port *port)
> struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> int ret;
>
> + if (port_priv->model != XR21V141X)
> + return xr_set_reg_uart(port,
> + xr_hal_table[port_priv->model][REG_ENABLE],
> + UART_ENABLE_TX | UART_ENABLE_RX);
> +

I handle this by overriding the default implementation for XR21V141X
instead.

> ret = xr_set_reg_um(port, UM_FIFO_ENABLE_REG,
> UM_ENABLE_TX_FIFO);
> if (ret)
> @@ -272,11 +355,37 @@ static int xr_uart_disable(struct usb_serial_port *port)
> if (ret)
> return ret;
>
> + if (port_priv->model != XR21V141X)
> + return 0;
> +

That allows not spreading conditionals like this throughout the driver.

> ret = xr_set_reg_um(port, UM_FIFO_ENABLE_REG, 0);
>
> return ret;
> }
>
> +static int fifo_reset(struct usb_serial_port *port)
> +{
> + struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> + int channel = port_priv->channel;
> + int ret = 0;
> +
> + if (port_priv->model != XR21V141X)
> + return 0;
> +
> + if (channel)
> + channel--;

This isn't very nice; a zero-based index is more natural.

> +
> + ret = xr_set_reg_um(port,
> + URM_RESET_RX_FIFO_BASE + channel, 0xff);
> + if (ret)
> + return ret;
> +
> + ret = xr_set_reg_um(port,
> + URM_RESET_TX_FIFO_BASE + channel, 0xff);
> +
> + return ret;
> +}

This looks malplaced, since it implements FIFO reset for the currently
supported types, but not for the one you're adding support for here.

> +
> static int xr_tiocmget(struct tty_struct *tty)
> {
> struct usb_serial_port *port = tty->driver_data;
> @@ -358,6 +467,12 @@ static void xr_break_ctl(struct tty_struct *tty, int break_state)
> struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> u8 state;
>
> + if (port_priv->model != XR21V141X) {
> + xr_usb_serial_ctrl_msg(port, USB_CDC_REQ_SEND_BREAK, state,
> + NULL, 0);
> + return;
> + }

This isn't needed at all since all types have TX_BREAK register.

> +
> if (break_state == 0)
> state = UART_BREAK_OFF;
> else
> @@ -505,6 +620,13 @@ static void xr_set_flow_mode(struct tty_struct *tty,
> flow = UART_FLOW_MODE_NONE;
> }
>
> + /*
> + * Add support for the TXT and RXT function for 0x1420, 0x1422, 0x1424,
> + * by setting GPIO_MODE [9:8] = '11'
> + */
> + if (port_priv->model == XR21B142X)
> + gpio_mode |= 0x300;

Not needed either, since we're reading back the (default) gpio mode just
above.

But again, you're not increasing the type width here, which is still u8
so this has no effect currently.

> +
> /*
> * As per the datasheet, UART needs to be disabled while writing to
> * FLOW_CONTROL register.
> @@ -521,9 +643,55 @@ static void xr_set_flow_mode(struct tty_struct *tty,
> xr_dtr_rts(port, 1);
> }
>
> -static void xr_set_termios(struct tty_struct *tty,
> - struct usb_serial_port *port,
> - struct ktermios *old_termios)
> +static void xr_set_termios_cdc(struct tty_struct *tty,
> + struct usb_serial_port *port,
> + struct ktermios *old_termios)
> +{
> + struct ktermios *termios = &tty->termios;
> + struct usb_cdc_line_coding line = { 0 };
> + int clear, set;
> +
> + line.dwDTERate = cpu_to_le32(tty_get_baud_rate(tty));
> + line.bCharFormat = termios->c_cflag & CSTOPB ? 1 : 0;
> + line.bParityType = termios->c_cflag & PARENB ?
> + (termios->c_cflag & PARODD ? 1 : 2) +
> + (termios->c_cflag & CMSPAR ? 2 : 0) : 0;

I know this comes from cdc-acm.c, but those ?: are still horrid. :)

> +
> + switch (C_CSIZE(tty)) {
> + case CS5:
> + line.bDataBits = 5;
> + break;
> + case CS6:
> + line.bDataBits = 6;
> + break;

The types you add later does not support CS5 and CS6 but IIRC you don't
handle that.

> + case CS7:
> + line.bDataBits = 7;
> + break;
> + case CS8:
> + default:
> + line.bDataBits = 8;
> + break;
> + }
> +
> + if (!line.dwDTERate) {
> + line.dwDTERate = tty->termios.c_ospeed;

This is broken, and still sets the line rate to 0.

> + clear = UART_MODE_DTR;
> + } else {
> + set = UART_MODE_DTR;
> + }
> +
> + if (clear || set)
> + xr_tiocmset_port(port, set, clear);

This is already handled by the driver in xr_set_flow_mode() below.

> + xr_set_flow_mode(tty, port, old_termios);
> +
> + xr_usb_serial_ctrl_msg(port, USB_CDC_REQ_SET_LINE_CODING, 0,
> + &line, sizeof(line));
> +}
> +
> +static void xr_set_termios_format_reg(struct tty_struct *tty,
> + struct usb_serial_port *port,
> + struct ktermios *old_termios)
> {
> struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> struct ktermios *termios = &tty->termios;
> @@ -532,6 +700,8 @@ static void xr_set_termios(struct tty_struct *tty,
> if (!old_termios || (tty->termios.c_ospeed != old_termios->c_ospeed))
> xr_set_baudrate(tty, port);
>
> + /* For models with a private CHARACTER_FORMAT register */
> +
> switch (C_CSIZE(tty)) {
> case CS5:
> case CS6:
> @@ -577,6 +747,28 @@ static void xr_set_termios(struct tty_struct *tty,
> xr_set_flow_mode(tty, port, old_termios);
> }
>
> +static void xr_set_termios(struct tty_struct *tty,
> + struct usb_serial_port *port,
> + struct ktermios *old_termios)
> +{
> + struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> +
> + /*
> + * Different models have different ways to setup character format:
> + *
> + * - XR2280X and XR21V141X have their on private register. On
> + * such models, 5-6 bits is not supported;
> + * - The other models use a standard CDC register.
> + *
> + * As we need to do different things with regards to 5-6 bits,
> + * the actual implementation is made on two different functions.
> + */
> + if (xr_hal_table[port_priv->model][REG_FORMAT] == VIA_CDC_REGISTER)
> + xr_set_termios_cdc(tty, port, old_termios);
> + else
> + xr_set_termios_format_reg(tty, port, old_termios);
> +}
> +
> static int xr_open(struct tty_struct *tty, struct usb_serial_port *port)
> {
> struct xr_port_private *port_priv = usb_get_serial_data(port->serial);
> @@ -596,6 +788,13 @@ static int xr_open(struct tty_struct *tty, struct usb_serial_port *port)
> gpio_dir = UART_MODE_DTR | UART_MODE_RTS;
> xr_set_reg_uart(port, xr_hal_table[port_priv->model][REG_GPIO_DIR], gpio_dir);
>
> + ret = fifo_reset(port);
> + if (ret) {
> + dev_err(&port->dev, "Failed to reset FIFO\n");
> + return ret;
> + }
> +
> +
> /* Setup termios */
> if (tty)
> xr_set_termios(tty, port, NULL);
> @@ -618,25 +817,39 @@ static void xr_close(struct usb_serial_port *port)
>
> static int xr_probe(struct usb_serial *serial, const struct usb_device_id *id)
> {
> + struct usb_driver *driver = serial->type->usb_driver;
> struct usb_interface *intf = serial->interface;
> struct usb_endpoint_descriptor *data_ep;
> + struct usb_device *udev = serial->dev;
> struct xr_port_private *port_priv;
> - int ifnum;
> + struct usb_interface *ctrl_intf;
> + int ifnum, ctrl_ifnum;
>
> /* Attach only data interfaces */
> ifnum = intf->cur_altsetting->desc.bInterfaceNumber;
> if (!(ifnum % 2))
> return -ENODEV;
>
> + /* Control interfaces are the even numbers */
> + ctrl_ifnum = ifnum - ifnum % 2;
> +
> port_priv = kzalloc(sizeof(*port_priv), GFP_KERNEL);
> if (!port_priv)
> return -ENOMEM;
>
> data_ep = &intf->cur_altsetting->endpoint[0].desc;'

Whoops, NULL-deref in case of a malicous device without endpoints.

> + ctrl_intf = usb_ifnum_to_if(udev, ctrl_ifnum);
>
> + port_priv->control_if = usb_get_intf(ctrl_intf);
> port_priv->model = id->driver_info;
> port_priv->channel = data_ep->bEndpointAddress;
>
> + /* Wake up control interface */
> + pm_suspend_ignore_children(&ctrl_intf->dev, false);
> + if (driver->supports_autosuspend)
> + pm_runtime_enable(&ctrl_intf->dev);
> + else
> + pm_runtime_set_active(&ctrl_intf->dev);

Now this is really nasty, but all goes away with the new multi-interface
support.

> usb_set_serial_data(serial, port_priv);
>
> return 0;
> @@ -645,6 +858,15 @@ static int xr_probe(struct usb_serial *serial, const struct usb_device_id *id)
> static void xr_disconnect(struct usb_serial *serial)
> {
> struct xr_port_private *port_priv = usb_get_serial_data(serial);
> + struct usb_driver *driver = serial->type->usb_driver;
> + struct usb_interface *ctrl_intf = port_priv->control_if;
> +
> + if (driver->supports_autosuspend)
> + pm_runtime_disable(&ctrl_intf->dev);
> +
> + pm_runtime_set_suspended(&ctrl_intf->dev);
> +
> + usb_put_intf(ctrl_intf);
>
> usb_set_serial_data(serial, 0);
> kfree(port_priv);
> @@ -654,6 +876,11 @@ static const struct usb_device_id id_table[] = {
> { USB_DEVICE(0x04e2, 0x1410), .driver_info = XR21V141X},
> { USB_DEVICE(0x04e2, 0x1412), .driver_info = XR21V141X},
> { USB_DEVICE(0x04e2, 0x1414), .driver_info = XR21V141X},
> +
> + { USB_DEVICE(0x04e2, 0x1420), .driver_info = XR21B142X},
> + { USB_DEVICE(0x04e2, 0x1422), .driver_info = XR21B142X},
> + { USB_DEVICE(0x04e2, 0x1424), .driver_info = XR21B142X},
> +
> { }
> };
> MODULE_DEVICE_TABLE(usb, id_table);

Johan

Next message: Xiongwei Song: "[PATCH v2] powerpc/traps: Enhance readability for trap types"
Previous message: Miroslav Benes: "Re: [PATCH v3 00/16] x86,objtool: Optimize !RETPOLINE"
In reply to: Mauro Carvalho Chehab: "[PATCH v2 4/7] USB: serial: xr: add support for XR21B142X devices"
Next in thread: Mauro Carvalho Chehab: "[PATCH v2 1/7] USB: serial: xr: simplify its namespace"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]