Re: [PATCH v12 1/1] serial: core: Start managing serial controllers to enable runtime PM

From: Chen-Yu Tsai
Date: Mon Jun 05 2023 - 07:34:26 EST


Hi,

On Mon, Jun 5, 2023 at 2:15 PM Tony Lindgren <tony@xxxxxxxxxxx> wrote:
>
> * Tony Lindgren <tony@xxxxxxxxxxx> [230603 06:35]:
> > * Tony Lindgren <tony@xxxxxxxxxxx> [230603 05:41]:
> > > I don't think 8250_mtk needs to do register access before and after the
> > > serial port registration, but if it does, then adding custom read/write
> > > functions can be done that do not rely on initialized port like
> > > serial_out().
> >
> > Oh but mtk8250_runtime_suspend() calls serial_in(up, MTK_UART_DEBUG0), so
> > yeah if that gets called before registration is complete it causes a NULL
> > pointer exception. If the serial_ctrl and serial_port devices do runtime
> > suspend before port registration completes, things will fail.
> >
> > Sounds like doing pm_runtime_resume_and_get() in mtk8250_probe() might
> > fix the issue. Still seems that adding a custom read function for
> > mtk8250_runtime_suspend() to use instead of calling serial_in() should
> > not be needed.
>
> Looking at this again, if serial8250_register_8250_port() fails, then
> mtk8250_runtime_suspend() would again try to access uninitialized port.
>
> Here's a better untested version of the patch to try.
>
> Regards,
>
> Tony
>
> 8< ---------------------------
> diff --git a/drivers/tty/serial/8250/8250_mtk.c b/drivers/tty/serial/8250/8250_mtk.c
> --- a/drivers/tty/serial/8250/8250_mtk.c
> +++ b/drivers/tty/serial/8250/8250_mtk.c
> @@ -57,6 +57,8 @@
> #define MTK_UART_XON1 40 /* I/O: Xon character 1 */
> #define MTK_UART_XOFF1 42 /* I/O: Xoff character 1 */
>
> +#define MTK_UART_REGSHIFT 2
> +
> #ifdef CONFIG_SERIAL_8250_DMA
> enum dma_rx_status {
> DMA_RX_START = 0,
> @@ -69,6 +71,7 @@ struct mtk8250_data {
> int line;
> unsigned int rx_pos;
> unsigned int clk_count;
> + void __iomem *membase;
> struct clk *uart_clk;
> struct clk *bus_clk;
> struct uart_8250_dma *dma;
> @@ -187,6 +190,17 @@ static void mtk8250_dma_enable(struct uart_8250_port *up)
> }
> #endif
>
> +/* Read and write for register access before and after port registration */
> +static u32 __maybe_unused mtk8250_read(struct mtk8250_data *data, u32 reg)
> +{
> + return readl(data->membase + (reg << MTK_UART_REGSHIFT));
> +}
> +
> +static void mtk8250_write(struct mtk8250_data *data, u32 reg, u32 val)
> +{
> + writel(val, data->membase + (reg << MTK_UART_REGSHIFT));
> +}
> +
> static int mtk8250_startup(struct uart_port *port)
> {
> #ifdef CONFIG_SERIAL_8250_DMA
> @@ -425,11 +439,10 @@ mtk8250_set_termios(struct uart_port *port, struct ktermios *termios,
> static int __maybe_unused mtk8250_runtime_suspend(struct device *dev)
> {
> struct mtk8250_data *data = dev_get_drvdata(dev);
> - struct uart_8250_port *up = serial8250_get_port(data->line);
>
> /* wait until UART in idle status */
> while
> - (serial_in(up, MTK_UART_DEBUG0));
> + (mtk8250_read(data, MTK_UART_DEBUG0));

I believe it still gets stuck here sometimes.

With your earlier patch, it could get through registering the port, and
the console would show

11002000.serial: ttyS0 at MMIO 0x11002000 (irq = 240, base_baud =
1625000) is a ST16650V2

for the console UART.

Angelo mentioned that we should be using SLEEP_REQ/SLEEP_ACK registers
in the MTK UART hardware.

I tried reworking it into your patch here, but it causes issues with the
UART-based Bluetooth on one of my devices. After the UART runtime suspends
and resumes, something is off and causes the transfers during Bluetooth
init to become corrupt.

I'll try some more stuff, but the existing code seems timing dependent.
If I add too many printk statements to the runtime suspend/resume
callbacks, things seem to work. One time I even ended up with broken
UARTs but otherwise booted up the system.

ChenYu

>
> if (data->clk_count == 0U) {
> dev_dbg(dev, "%s clock count is 0\n", __func__);
> @@ -553,6 +566,7 @@ static int mtk8250_probe(struct platform_device *pdev)
> if (!data)
> return -ENOMEM;
>
> + data->membase = uart.port.membase;
> data->clk_count = 0;
>
> if (pdev->dev.of_node) {
> @@ -570,7 +584,7 @@ static int mtk8250_probe(struct platform_device *pdev)
> uart.port.flags = UPF_BOOT_AUTOCONF | UPF_FIXED_PORT;
> uart.port.dev = &pdev->dev;
> uart.port.iotype = UPIO_MEM32;
> - uart.port.regshift = 2;
> + uart.port.regshift = MTK_UART_REGSHIFT;
> uart.port.private_data = data;
> uart.port.shutdown = mtk8250_shutdown;
> uart.port.startup = mtk8250_startup;
> @@ -581,27 +595,30 @@ static int mtk8250_probe(struct platform_device *pdev)
> uart.dma = data->dma;
> #endif
>
> - /* Disable Rate Fix function */
> - writel(0x0, uart.port.membase +
> - (MTK_UART_RATE_FIX << uart.port.regshift));
> -
> platform_set_drvdata(pdev, data);
>
> pm_runtime_enable(&pdev->dev);
> - err = mtk8250_runtime_resume(&pdev->dev);
> + err = pm_runtime_resume_and_get(&pdev->dev);
> if (err)
> goto err_pm_disable;
>
> + /* Disable Rate Fix function */
> + mtk8250_write(data, 0, MTK_UART_RATE_FIX);
> +
> data->line = serial8250_register_8250_port(&uart);
> if (data->line < 0) {
> err = data->line;
> - goto err_pm_disable;
> + goto err_pm_put;
> }
>
> data->rx_wakeup_irq = platform_get_irq_optional(pdev, 1);
>
> + pm_runtime_put_sync(&pdev->dev);
> +
> return 0;
>
> +err_pm_put:
> + pm_runtime_put_sync(&pdev->dev);
> err_pm_disable:
> pm_runtime_disable(&pdev->dev);
>
> @@ -694,7 +711,7 @@ static int __init early_mtk8250_setup(struct earlycon_device *device,
> return -ENODEV;
>
> device->port.iotype = UPIO_MEM32;
> - device->port.regshift = 2;
> + device->port.regshift = MTK_UART_REGSHIFT;
>
> return early_serial8250_setup(device, NULL);
> }