Re: [PATCH AUTOSEL 6.18-6.6] serial: qcom_geni: Disable closing_wait for console to prevent shutdown timeout

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] serial: qcom_geni: uncancelled console TX in qcom_geni_serial_startup()
--

commit 410ad55ea7c02110e68e77ae15503cb551a22bf5
Author: Aniket Randive <aniket.randive@oss.qualcomm.com>

serial: qcom_geni: Disable closing_wait for console to prevent shutdown timeout

This commit sets closing_wait to ASYNC_CLOSING_WAIT_NONE in
qcom_geni_serial_startup() for console ports so that closing the console
TTY bypasses tty_wait_until_sent() during system shutdown.

Link: https://patch.msgid.link/20260706093749.836121-1-aniket.randive@oss.qualcomm.com

> diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c
> index 8058b839b26ce..219ac88b0fab2 100644
> --- a/drivers/tty/serial/qcom_geni_serial.c
> +++ b/drivers/tty/serial/qcom_geni_serial.c
> @@ -1227,6 +1227,7 @@ static int qcom_geni_serial_startup(struct uart_port *uport)
>  {
>  	int ret;
>  	struct qcom_geni_serial_port *port = to_dev_port(uport);
> +	struct tty_port *tport = &uport->state->port;
>  
>  	if (!port->setup) {
>  		ret = qcom_geni_serial_port_setup(uport);
> @@ -1234,6 +1235,13 @@ static int qcom_geni_serial_startup(struct uart_port *uport)
>  			return ret;
>  	}
>  
> +	/*
> +	 * Skip the close-time transmit drain for console ports so that
> +	 * shutdown can proceed without waiting for pending TX completion.
> +	 */
> +	if (uart_console(uport))
> +		tport->closing_wait = ASYNC_CLOSING_WAIT_NONE;

[Severity: Medium]
Does uart_shutdown() actually get called to cancel pending TX commands when a
console port is closed?

The commit message states that setting closing_wait to ASYNC_CLOSING_WAIT_NONE
bypasses tty_wait_until_sent() on close, allowing uart_shutdown() to cancel TX
commands cleanly.

However, when uart_console(uport) is true, serial_core_add_one_port() sets
port->console:

drivers/tty/serial/serial_core.c:serial_core_add_one_port() {
	...
	port->console = uart_console(uport);
	...
}

When userspace closes the final file descriptor on the console TTY via
tty_release() -> uart_close() -> tty_port_close(), tty_port_close_start()
skips tty_wait_until_sent() without calling tty_driver_flush_buffer() (unless
tty->flow.tco_stopped is set):

drivers/tty/tty_port.c:tty_port_close_start() {
	...
		if (tty->flow.tco_stopped)
			tty_driver_flush_buffer(tty);
		if (port->closing_wait != ASYNC_CLOSING_WAIT_NONE)
			tty_wait_until_sent(tty, port->closing_wait);
	...
}

Then tty_port_close() calls tty_port_shutdown(), which returns early for
console ports without invoking port->ops->shutdown() (uart_tty_port_shutdown()
-> uart_port_shutdown() -> qcom_geni_serial_shutdown()):

drivers/tty/tty_port.c:tty_port_shutdown() {
	struct tty_struct *tty = port->tty;

	mutex_lock(&port->mutex);
	if (port->console)
		goto out;
	...
}

After that, tty_port_close_end() clears TTY_PORT_ACTIVE via
tty_port_set_active(port, false), so even a subsequent uart_hangup() (which
checks tty_port_active(port)) will not call uart_flush_buffer() or
uart_shutdown().

Doesn't this leave pending GENI sequencer TX commands (UART_START_TX), TX
interrupts, and unsent bytes in tport->xmit_fifo active after the console TTY
is closed, causing subsequent qcom_geni_serial_console_write() calls to poll
until timeout before cancelling the stalled command and re-triggering
qcom_geni_serial_isr() -> qcom_geni_serial_handle_tx_fifo() on the leftover
xmit_fifo data?

> +
>  	uart_port_lock_irq(uport);
>  	qcom_geni_serial_start_rx(uport);
>  	uart_port_unlock_irq(uport);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261019135441.1420968-515-sashal@kernel.org?part=1