Re: [PATCH 2/5] serial: imx: Remove redundant dev_err()

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Pre-existing issues:
- [High] The global `imx_uart_ports` array retains a dangling pointer to a freed `sport` structure if `uart_add_one_port()` fails or when the driver is unbound.
- [High] The hrtimers `trigger_start_tx` and `trigger_stop_tx` are never synchronously cancelled during device teardown, leading to a potential Use-After-Free.
--

--- Patch [2]: [PATCH 2/5] serial: imx: Remove redundant dev_err() ---
commit 44174bba1c656c63bfedf3dc6b8ca0b3fa1165cb
Author: Pan Chuang <[email protected]>

serial: imx: Remove redundant dev_err()

This commit removes driver-specific dev_err() calls after devm_request_irq(), 
as the core function already logs detailed error messages on failure since 
commit 55b48e23f5c4.

> diff --git a/drivers/tty/serial/imx.c b/drivers/tty/serial/imx.c
> index 251a50c8aa387..95fdb60e06a77 100644
> --- a/drivers/tty/serial/imx.c
> +++ b/drivers/tty/serial/imx.c
> @@ -2602,34 +2602,23 @@ static int imx_uart_probe(struct platform_device *pdev)
[ ... ]
>  	} else {
>  		ret = devm_request_irq(&pdev->dev, rxirq, imx_uart_int, 0,
>  				       dev_name(&pdev->dev), sport);
> -		if (ret) {
> -			dev_err(&pdev->dev, "failed to request irq: %d\n", ret);
> +		if (ret)
>  			goto err_clk;
> -		}
>  	}
>  
>  	imx_uart_ports[sport->port.line] = sport;

[Severity: High]
This is a pre-existing issue, but does this leave a dangling pointer in
the global array if uart_add_one_port() fails later in the probe function?

If uart_add_one_port() fails, the function returns an error and devm
automatically frees the sport structure. However, the pointer was already
assigned to imx_uart_ports[sport->port.line] and is never cleared on the
error path. Could this lead to a use-after-free if the console subsystem or
sysrq accesses the array?

Similarly, it looks like imx_uart_remove() does not clear this array entry
before devres frees the memory during driver unbind.

[Severity: High]
This is also a pre-existing issue, but are the trigger_start_tx and
trigger_stop_tx hrtimers synchronously cancelled during device teardown?

Looking at the unbind path in imx_uart_remove():

static void imx_uart_remove(struct platform_device *pdev)
{
        struct imx_port *sport = platform_get_drvdata(pdev);

        uart_remove_one_port(&imx_uart_uart_driver, &sport->port);
}

During unbind, imx_uart_shutdown() tries to stop the port and calls
hrtimer_try_to_cancel() implicitly. However, hrtimer_try_to_cancel() doesn't
wait for a currently executing callback to finish.

Since explicit hrtimer_cancel() is never called in imx_uart_remove(),
could devres free the sport structure while the hrtimer callback is still
running or pending on another CPU, resulting in a use-after-free?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2