Re: [PATCH RFC v2 07/10] tty: serial: Add Cortina UART driver

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] tty: serial: use-after-free due to __init on console setup
--

commit f274bba502abaeb866daa29a51921a3960bdaaa5
Author: Fil Dunsky <filipp.dunsky@gmail.com>

tty: serial: Add Cortina UART driver

This commit adds a driver for the UART found in Cortina SoCs. It supports
the system console and earlycon, using the current serial core API and devm
resource management.

> diff --git a/drivers/tty/serial/cortina-uart.c b/drivers/tty/serial/cortina-uart.c
> new file mode 100644
> index 0000000000000..1952629de0522
> --- /dev/null
> +++ b/drivers/tty/serial/cortina-uart.c
> @@ -0,0 +1,583 @@

[ ... ]

> +static int __init cortina_uart_console_setup(struct console *co, char *options)
> +{
> +	struct cortina_uart_port *up;

[Severity: High]
Is it safe to mark this function with __init?

Since cortina_uart_probe() uses devm_clk_get_enabled() which can return
-EPROBE_DEFER, the platform device might probe long after the boot process
has finished and the .init.text section has been freed.

During a post-boot probe, uart_add_one_port() invokes register_console(),
which unconditionally calls the console's .setup callback. If
cortina_uart_console_setup() is marked as __init, jumping to it at this
point would cause a kernel panic due to executing freed memory.

[ ... ]

> +static int cortina_uart_probe(struct platform_device *pdev)
> +{
> +	struct device *dev = &pdev->dev;

[ ... ]

> +	up->clk = devm_clk_get_enabled(dev, NULL);
> +	if (IS_ERR(up->clk))
> +		return dev_err_probe(dev, PTR_ERR(up->clk), "failed to get clock\n");

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261006-b4-cs75xx-v2-0-b7d6e1e24d44@gmail.com?part=7