[PATCH tty v4 0/6] 8250: Add console flow control

[PATCH tty v4 0/6] 8250: Add console flow control

[John Ogness]

Hi,

This is v4 of a series to implement console flow control for the
8250 serial driver. v3 is here [0].

The 8250 driver already has code in place to support console flow
control. However, there is no way to activate it and it is
incomplete. This series provides the necessary missing pieces while
attempting to be as conservative as possible, so as not to introduce
any side effects into the many 8250 variants or other non-8250 serial
drivers.

For patch 2 I used the following Coccinelle script to perform the
modifications...

===== BEGIN cons_flow.cocci =====
// SPDX-License-Identifier: GPL-2.0-only
// Options: --all-includes

virtual patch

@r1@
type T1;
identifier U;
@@

T1 {
   ...
   struct uart_port U;
   ...
};

@r2@
r1.T1 *E;
@@

- (E->port.flags & UPF_CONS_FLOW)
+ uart_cons_flow_enabled(&E->port)

@r3@
struct uart_port *U;
@@

- (U->flags & UPF_CONS_FLOW)
+ uart_cons_flow_enabled(U)

@r4@
struct uart_port *U;
@@

- U->flags |= UPF_CONS_FLOW
+ uart_set_cons_flow_enabled(U, true)
===== END cons_flow.cocci =====

Changes for v4:

- Rename uart_get_cons_flow() to uart_cons_flow_enabled().

- Rename uart_set_cons_flow() to uart_set_cons_flow_enabled().

Changes for v3:

- Deprecate UPF_CONS_FLOW. Provide separate boolean with wrappers as
  alternative.

- Update all UPF_CONS_FLOW users to new cons_flow wrappers.

- Use irqsave variant of spin lock for status update.

- When 8250 console flow control is not specified, clear the policy.

Changes for v2:

- Prepend a patch to perform an extra LSR wait after CTS assertion if
  the initial LSR wait timed out.

- Close a window in serial8250_register_8250_port() where console
  flow control was briefly disabled.

- Add port lock synchronization to the port->status RMW update in
  uart_set_options().

John Ogness

[0] https://lore.kernel.org/lkml/20260417102423.40984-1-john.ogness@linutronix.de

John Ogness (6):
  serial: core: Add dedicated uart_port field for console flow
  serial: Replace driver usage of UPF_CONS_FLOW
  serial: sh-sci: Avoid deprecated UPF_CONS_FLOW
  serial: 8250: Set cons_flow on port registration
  serial: 8250: Check LSR timeout on console flow control
  serial: 8250: Add support for console flow control

 drivers/tty/serial/8250/8250_core.c |  6 ++++++
 drivers/tty/serial/8250/8250_port.c | 21 +++++++++++++++++----
 drivers/tty/serial/bcm63xx_uart.c   |  2 +-
 drivers/tty/serial/omap-serial.c    |  2 +-
 drivers/tty/serial/pch_uart.c       |  2 +-
 drivers/tty/serial/pxa.c            |  2 +-
 drivers/tty/serial/samsung_tty.c    |  8 ++++----
 drivers/tty/serial/serial_core.c    | 21 ++++++++++++++++++++-
 drivers/tty/serial/serial_txx9.c    |  4 ++--
 drivers/tty/serial/sh-sci.c         |  5 ++++-
 drivers/tty/serial/sunsu.c          |  2 +-
 include/linux/serial_core.h         | 20 ++++++++++++++++++++
 12 files changed, 78 insertions(+), 17 deletions(-)


base-commit: a2083fd1fa7aa0ef5cd8fd92396da0de2d0654b0
-- 
2.47.3

[PATCH tty v4 1/6] serial: core: Add dedicated uart_port field for console flow

[John Ogness]

Currently the UPF_CONS_FLOW bit in the uart_port.flags field is used
by serial console drivers to identify if a user has configured flow
control on the console. Usually this policy is setup during early
boot, but can be changed at runtime.

The bits in uart_port.flags are either hardware and driver
properties that are initialized before usage or are properties that
can be changed via the tty layer.

The UPF_CONS_FLOW bit is an exception because it is a console-only
policy that can change at runtime and its setting and usage have
nothing to do with the tty layer. This actually causes a problem
for its usage because uart_port.flags is synchronized by a related
tty_port.mutex, but a console has no relation to a tty (other than
sharing the port).

This is probably why console flow control is not properly available
for most serial drivers. And it is hindering being able to provide a
proper implementation. Commit d01f4d181c92 ("serial: core: Privatize
tty->hw_stopped") addressed a similar issue to deal with software
assisted CTS flow state tracking.

Add a new uart_port boolean field "cons_flow" to store the user
configuration for console flow control. Add getter/setter wrappers
to allow for adding more policies later and/or locking constraint
validation.

Mark UPF_CONS_FLOW as deprecated.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 include/linux/serial_core.h | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h
index 666430b478997..4f7bbdd900176 100644
--- a/include/linux/serial_core.h
+++ b/include/linux/serial_core.h
@@ -533,6 +533,7 @@ struct uart_port {
 #define UPF_HARD_FLOW		((__force upf_t) (UPF_AUTO_CTS | UPF_AUTO_RTS))
 /* Port has hardware-assisted s/w flow control */
 #define UPF_SOFT_FLOW		((__force upf_t) BIT_ULL(22))
+/* Deprecated: use uart_set_cons_flow_enabled()/uart_cons_flow_enabled() instead. */
 #define UPF_CONS_FLOW		((__force upf_t) BIT_ULL(23))
 #define UPF_SHARE_IRQ		((__force upf_t) BIT_ULL(24))
 #define UPF_EXAR_EFR		((__force upf_t) BIT_ULL(25))
@@ -567,6 +568,7 @@ struct uart_port {
 #define UPSTAT_SYNC_FIFO	((__force upstat_t) (1 << 5))
 
 	bool			hw_stopped;		/* sw-assisted CTS flow state */
+	bool			cons_flow;		/* user specified console flow control */
 	unsigned int		mctrl;			/* current modem ctrl settings */
 	unsigned int		frame_time;		/* frame timing in ns */
 	unsigned int		type;			/* port type */
@@ -1163,6 +1165,16 @@ static inline bool uart_softcts_mode(struct uart_port *uport)
 	return ((uport->status & mask) == UPSTAT_CTS_ENABLE);
 }
 
+static inline void uart_set_cons_flow_enabled(struct uart_port *uport, bool enabled)
+{
+	uport->cons_flow = enabled;
+}
+
+static inline bool uart_cons_flow_enabled(const struct uart_port *uport)
+{
+	return uport->cons_flow;
+}
+
 /*
  * The following are helper functions for the low level drivers.
  */
-- 
2.47.3

[PATCH tty v4 2/6] serial: Replace driver usage of UPF_CONS_FLOW

[John Ogness]

Rather than using the UPF_CONS_FLOW bit of uart_port.flags to track
the user configuration of console flow control, use the newly added
uart_port.cons_flow (via its getter/setter functions).

A coccinelle script was used to perform the search/replace.

Note1: The sh-sci driver is blindly copying platform data configuration
       flags to uart_port.flags. Thus UPF_CONS_FLOW could get set for
       uart_port.flags. A follow-up commit will address this.

Note2: The samsung_tty driver is using UPF_CONS_FLOW as a platform data
       configuration flag. However, the driver explicitly checks for
       this configuration flag and thus setting UPF_CONS_FLOW in
       uart_port.flags is avoided.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/tty/serial/8250/8250_port.c | 4 ++--
 drivers/tty/serial/bcm63xx_uart.c   | 2 +-
 drivers/tty/serial/omap-serial.c    | 2 +-
 drivers/tty/serial/pch_uart.c       | 2 +-
 drivers/tty/serial/pxa.c            | 2 +-
 drivers/tty/serial/samsung_tty.c    | 8 ++++----
 drivers/tty/serial/serial_txx9.c    | 4 ++--
 drivers/tty/serial/sunsu.c          | 2 +-
 8 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
index af78cc02f38e7..dbed5e5767541 100644
--- a/drivers/tty/serial/8250/8250_port.c
+++ b/drivers/tty/serial/8250/8250_port.c
@@ -1988,7 +1988,7 @@ static void wait_for_xmitr(struct uart_8250_port *up, int bits)
 	wait_for_lsr(up, bits);
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->port.flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(&up->port)) {
 		for (tmout = 1000000; tmout; tmout--) {
 			unsigned int msr = serial_in(up, UART_MSR);
 			up->msr_saved_flags |= msr & MSR_SAVE_FLAGS;
@@ -3351,7 +3351,7 @@ void serial8250_console_write(struct uart_8250_port *up, const char *s,
 		 * it regardless of the CTS state. Therefore, only use fifo
 		 * if we don't use control flow.
 		 */
-		!(up->port.flags & UPF_CONS_FLOW);
+		!uart_cons_flow_enabled(&up->port);
 
 	if (likely(use_fifo))
 		serial8250_console_fifo_write(up, s, count);
diff --git a/drivers/tty/serial/bcm63xx_uart.c b/drivers/tty/serial/bcm63xx_uart.c
index 51df9d2d8bfc5..544695cb184c3 100644
--- a/drivers/tty/serial/bcm63xx_uart.c
+++ b/drivers/tty/serial/bcm63xx_uart.c
@@ -675,7 +675,7 @@ static void wait_for_xmitr(struct uart_port *port)
 	}
 
 	/* Wait up to 1s for flow control if necessary */
-	if (port->flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(port)) {
 		tmout = 1000000;
 		while (--tmout) {
 			unsigned int val;
diff --git a/drivers/tty/serial/omap-serial.c b/drivers/tty/serial/omap-serial.c
index 0b85f47ff19e0..a689d190940cf 100644
--- a/drivers/tty/serial/omap-serial.c
+++ b/drivers/tty/serial/omap-serial.c
@@ -1092,7 +1092,7 @@ static void __maybe_unused wait_for_xmitr(struct uart_omap_port *up)
 	} while (!uart_lsr_tx_empty(status));
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->port.flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(&up->port)) {
 		for (tmout = 1000000; tmout; tmout--) {
 			unsigned int msr = serial_in(up, UART_MSR);
 
diff --git a/drivers/tty/serial/pch_uart.c b/drivers/tty/serial/pch_uart.c
index 6729d8e83c3c5..2f72cd62e2488 100644
--- a/drivers/tty/serial/pch_uart.c
+++ b/drivers/tty/serial/pch_uart.c
@@ -1444,7 +1444,7 @@ static void wait_for_xmitr(struct eg20t_port *up, int bits)
 	}
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->port.flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(&up->port)) {
 		unsigned int tmout;
 		for (tmout = 1000000; tmout; tmout--) {
 			unsigned int msr = ioread8(up->membase + UART_MSR);
diff --git a/drivers/tty/serial/pxa.c b/drivers/tty/serial/pxa.c
index fea0255067ccd..10fc8990579b5 100644
--- a/drivers/tty/serial/pxa.c
+++ b/drivers/tty/serial/pxa.c
@@ -573,7 +573,7 @@ static void wait_for_xmitr(struct uart_pxa_port *up)
 	} while (!uart_lsr_tx_empty(status));
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->port.flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(&up->port)) {
 		tmout = 1000000;
 		while (--tmout &&
 		       ((serial_in(up, UART_MSR) & UART_MSR_CTS) == 0))
diff --git a/drivers/tty/serial/samsung_tty.c b/drivers/tty/serial/samsung_tty.c
index e27806bf2cf3e..2f94fc798cffb 100644
--- a/drivers/tty/serial/samsung_tty.c
+++ b/drivers/tty/serial/samsung_tty.c
@@ -319,7 +319,7 @@ static void s3c24xx_serial_stop_tx(struct uart_port *port)
 	ourport->tx_enabled = 0;
 	ourport->tx_in_progress = 0;
 
-	if (port->flags & UPF_CONS_FLOW)
+	if (uart_cons_flow_enabled(port))
 		s3c24xx_serial_rx_enable(port);
 
 	ourport->tx_mode = 0;
@@ -493,7 +493,7 @@ static void s3c24xx_serial_start_tx(struct uart_port *port)
 	struct tty_port *tport = &port->state->port;
 
 	if (!ourport->tx_enabled) {
-		if (port->flags & UPF_CONS_FLOW)
+		if (uart_cons_flow_enabled(port))
 			s3c24xx_serial_rx_disable(port);
 
 		ourport->tx_enabled = 1;
@@ -781,7 +781,7 @@ static void s3c24xx_serial_rx_drain_fifo(struct s3c24xx_uart_port *ourport)
 		uerstat = rd_regl(port, S3C2410_UERSTAT);
 		ch = rd_reg(port, S3C2410_URXH);
 
-		if (port->flags & UPF_CONS_FLOW) {
+		if (uart_cons_flow_enabled(port)) {
 			bool txe = s3c24xx_serial_txempty_nofifo(port);
 
 			if (ourport->rx_enabled) {
@@ -1830,7 +1830,7 @@ static int s3c24xx_serial_init_port(struct s3c24xx_uart_port *ourport,
 
 	if (cfg->uart_flags & UPF_CONS_FLOW) {
 		dev_dbg(port->dev, "enabling flow control\n");
-		port->flags |= UPF_CONS_FLOW;
+		uart_set_cons_flow_enabled(port, true);
 	}
 
 	/* sort our the physical and virtual addresses for each UART */
diff --git a/drivers/tty/serial/serial_txx9.c b/drivers/tty/serial/serial_txx9.c
index 436a559234dfe..4ae9a45c8e3a3 100644
--- a/drivers/tty/serial/serial_txx9.c
+++ b/drivers/tty/serial/serial_txx9.c
@@ -422,7 +422,7 @@ static void wait_for_xmitr(struct uart_port *up)
 		udelay(1);
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(up)) {
 		tmout = 1000000;
 		while (--tmout &&
 		       (sio_in(up, TXX9_SICISR) & TXX9_SICISR_CTSS))
@@ -857,7 +857,7 @@ serial_txx9_console_write(struct console *co, const char *s, unsigned int count)
 	 *	Disable flow-control if enabled (and unnecessary)
 	 */
 	flcr = sio_in(up, TXX9_SIFLCR);
-	if (!(up->flags & UPF_CONS_FLOW) && (flcr & TXX9_SIFLCR_TES))
+	if (!uart_cons_flow_enabled(up) && (flcr & TXX9_SIFLCR_TES))
 		sio_out(up, TXX9_SIFLCR, flcr & ~TXX9_SIFLCR_TES);
 
 	uart_console_write(up, s, count, serial_txx9_console_putchar);
diff --git a/drivers/tty/serial/sunsu.c b/drivers/tty/serial/sunsu.c
index 6505a1930da9a..4ff3e5c41dab9 100644
--- a/drivers/tty/serial/sunsu.c
+++ b/drivers/tty/serial/sunsu.c
@@ -1245,7 +1245,7 @@ static void wait_for_xmitr(struct uart_sunsu_port *up)
 	} while (!uart_lsr_tx_empty(status));
 
 	/* Wait up to 1s for flow control if necessary */
-	if (up->port.flags & UPF_CONS_FLOW) {
+	if (uart_cons_flow_enabled(&up->port)) {
 		tmout = 1000000;
 		while (--tmout &&
 		       ((serial_in(up, UART_MSR) & UART_MSR_CTS) == 0))
-- 
2.47.3

[PATCH tty v4 3/6] serial: sh-sci: Avoid deprecated UPF_CONS_FLOW

[John Ogness]

Avoid setting the uart_port.flags deprecated UPF_CONS_FLOW bit if it
has been configured in the platform data. Use the new cons_flow
wrappers instead.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/tty/serial/sh-sci.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
index 6c819b6b24258..a1328a3b6638d 100644
--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -3369,9 +3369,12 @@ static int sci_init_single(struct platform_device *dev,
 	}
 
 	port->type		= SCI_PUBLIC_PORT_ID(p->type);
-	port->flags		= UPF_FIXED_PORT | UPF_BOOT_AUTOCONF | p->flags;
+	port->flags		= UPF_FIXED_PORT | UPF_BOOT_AUTOCONF |
+				  (p->flags & ~UPF_CONS_FLOW);
 	port->fifosize		= sci_port->params->fifosize;
 
+	uart_set_cons_flow_enabled(port, p->flags & UPF_CONS_FLOW);
+
 	if (p->type == PORT_SCI && !dev->dev.of_node) {
 		if (sci_port->reg_size >= 0x20)
 			port->regshift = 2;
-- 
2.47.3

[PATCH tty v4 4/6] serial: 8250: Set cons_flow on port registration

[John Ogness]

Since console flow control policy is no longer part of uart_port.flags,
explicitly set the policy for the port.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/tty/serial/8250/8250_core.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/tty/serial/8250/8250_core.c b/drivers/tty/serial/8250/8250_core.c
index a428e88938eb7..b0275204e1167 100644
--- a/drivers/tty/serial/8250/8250_core.c
+++ b/drivers/tty/serial/8250/8250_core.c
@@ -746,6 +746,8 @@ int serial8250_register_8250_port(const struct uart_8250_port *up)
 	uart->lsr_save_mask	= up->lsr_save_mask;
 	uart->dma		= up->dma;
 
+	uart_set_cons_flow_enabled(&uart->port, uart_cons_flow_enabled(&up->port));
+
 	/* Take tx_loadsz from fifosize if it wasn't set separately */
 	if (uart->port.fifosize && !uart->tx_loadsz)
 		uart->tx_loadsz = uart->port.fifosize;
-- 
2.47.3

[PATCH tty v4 5/6] serial: 8250: Check LSR timeout on console flow control

[John Ogness]

wait_for_xmitr() calls wait_for_lsr() to wait for the transmission
registers to be empty. wait_for_lsr() can timeout after a reasonable
amount of time.

When console flow control is active, wait_for_xmitr() additionally
polls CTS, waiting for the peer to signal that it is ready to receive
more data.

If hardware flow control is enabled (auto CTS) and the peer deasserts
CTS, wait_for_lsr() will timeout. If additionally console flow
control is active and while polling CTS the peer asserts CTS, the
console will assume it can immediately transmit, even though the
transmission registers may not be empty. This can lead to data loss.

Avoid this problem by performing an extra wait_for_lsr() upon CTS
assertion if wait_for_lsr() previously timed out.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/tty/serial/8250/8250_port.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
index dbed5e5767541..719789013f6df 100644
--- a/drivers/tty/serial/8250/8250_port.c
+++ b/drivers/tty/serial/8250/8250_port.c
@@ -1984,16 +1984,20 @@ static bool wait_for_lsr(struct uart_8250_port *up, int bits)
 static void wait_for_xmitr(struct uart_8250_port *up, int bits)
 {
 	unsigned int tmout;
+	bool tx_ready;
 
-	wait_for_lsr(up, bits);
+	tx_ready = wait_for_lsr(up, bits);
 
 	/* Wait up to 1s for flow control if necessary */
 	if (uart_cons_flow_enabled(&up->port)) {
 		for (tmout = 1000000; tmout; tmout--) {
 			unsigned int msr = serial_in(up, UART_MSR);
 			up->msr_saved_flags |= msr & MSR_SAVE_FLAGS;
-			if (msr & UART_MSR_CTS)
+			if (msr & UART_MSR_CTS) {
+				if (!tx_ready)
+					wait_for_lsr(up, bits);
 				break;
+			}
 			udelay(1);
 			touch_nmi_watchdog();
 		}
-- 
2.47.3

[PATCH tty v4 6/6] serial: 8250: Add support for console flow control

[John Ogness]

The kernel documentation specifies that the console option 'r' can
be used to enable hardware flow control for console writes. The 8250
driver does include code for hardware flow control on the console if
cons_flow is set, but there is no code path that actually sets this.
However, that is not the only issue. The problems are:

1. Specifying the console option 'r' does not lead to cons_flow being
   set.

2. Even if cons_flow would be set, serial8250_register_8250_port()
   clears it.

3. When the console option 'r' is specified, uart_set_options()
   attempts to initialize the port for CRTSCTS. However, afterwards
   it does not set the UPSTAT_CTS_ENABLE status bit and therefore on
   boot, uart_cts_enabled() is always false. This policy bit is
   important for console drivers as a criteria if they may poll CTS.

4. Even though uart_set_options() attempts to initialize the port
   for CRTSCTS, the 8250 set_termios() callback does not enable the
   RTS signal (TIOCM_RTS) and thus the hardware is not properly
   initialized for CTS polling.

5. Even if modem control was properly setup for CTS polling
   (TIOCM_RTS), uart_configure_port() clears TIOCM_RTS, thus
   breaking CTS polling.

6. wait_for_xmitr() and serial8250_console_write() use cons_flow
   to decide if CTS polling should occur. However, the condition
   should also include a check that it is not in RS485 mode and
   CRTSCTS is actually enabled in the hardware.

Address all these issues as conservatively as possible by gating them
behind checks focussed on the user specifying console hardware flow
control support and the hardware being configured for CTS polling
at the time of the write to the UART.

Since checking the UPSTAT_CTS_ENABLE status bit is a part of the new
condition gate, these changes also support runtime termios updates to
disable/enable CRTSCTS.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/tty/serial/8250/8250_core.c |  6 +++++-
 drivers/tty/serial/8250/8250_port.c | 13 +++++++++++--
 drivers/tty/serial/serial_core.c    | 21 ++++++++++++++++++++-
 include/linux/serial_core.h         |  8 ++++++++
 4 files changed, 44 insertions(+), 4 deletions(-)

diff --git a/drivers/tty/serial/8250/8250_core.c b/drivers/tty/serial/8250/8250_core.c
index b0275204e1167..1f03da85e3414 100644
--- a/drivers/tty/serial/8250/8250_core.c
+++ b/drivers/tty/serial/8250/8250_core.c
@@ -693,6 +693,7 @@ static void serial_8250_overrun_backoff_work(struct work_struct *work)
 int serial8250_register_8250_port(const struct uart_8250_port *up)
 {
 	struct uart_8250_port *uart;
+	bool cons_flow;
 	int ret;
 
 	if (up->port.uartclk == 0)
@@ -716,6 +717,9 @@ int serial8250_register_8250_port(const struct uart_8250_port *up)
 	if (uart->port.type == PORT_8250_CIR)
 		return -ENODEV;
 
+	/* Preserve specified console flow control. */
+	cons_flow = uart_cons_flow_enabled(&uart->port);
+
 	if (uart->port.dev)
 		uart_remove_one_port(&serial8250_reg, &uart->port);
 
@@ -746,7 +750,7 @@ int serial8250_register_8250_port(const struct uart_8250_port *up)
 	uart->lsr_save_mask	= up->lsr_save_mask;
 	uart->dma		= up->dma;
 
-	uart_set_cons_flow_enabled(&uart->port, uart_cons_flow_enabled(&up->port));
+	uart_set_cons_flow_enabled(&uart->port, uart_cons_flow_enabled(&up->port) | cons_flow);
 
 	/* Take tx_loadsz from fifosize if it wasn't set separately */
 	if (uart->port.fifosize && !uart->tx_loadsz)
diff --git a/drivers/tty/serial/8250/8250_port.c b/drivers/tty/serial/8250/8250_port.c
index 719789013f6df..9ba3876bffa50 100644
--- a/drivers/tty/serial/8250/8250_port.c
+++ b/drivers/tty/serial/8250/8250_port.c
@@ -1989,7 +1989,7 @@ static void wait_for_xmitr(struct uart_8250_port *up, int bits)
 	tx_ready = wait_for_lsr(up, bits);
 
 	/* Wait up to 1s for flow control if necessary */
-	if (uart_cons_flow_enabled(&up->port)) {
+	if (uart_console_hwflow_active(&up->port)) {
 		for (tmout = 1000000; tmout; tmout--) {
 			unsigned int msr = serial_in(up, UART_MSR);
 			up->msr_saved_flags |= msr & MSR_SAVE_FLAGS;
@@ -2786,6 +2786,12 @@ serial8250_do_set_termios(struct uart_port *port, struct ktermios *termios,
 		serial8250_set_efr(port, termios);
 		serial8250_set_divisor(port, baud, quot, frac);
 		serial8250_set_fcr(port, termios);
+		/* Consoles manually poll CTS for hardware flow control. */
+		if (uart_console(port) &&
+		    !(port->rs485.flags & SER_RS485_ENABLED)
+		    && termios->c_cflag & CRTSCTS) {
+			port->mctrl |= TIOCM_RTS;
+		}
 		serial8250_set_mctrl(port, port->mctrl);
 	}
 
@@ -3355,7 +3361,7 @@ void serial8250_console_write(struct uart_8250_port *up, const char *s,
 		 * it regardless of the CTS state. Therefore, only use fifo
 		 * if we don't use control flow.
 		 */
-		!uart_cons_flow_enabled(&up->port);
+		!uart_console_hwflow_active(&up->port);
 
 	if (likely(use_fifo))
 		serial8250_console_fifo_write(up, s, count);
@@ -3425,6 +3431,9 @@ int serial8250_console_setup(struct uart_port *port, char *options, bool probe)
 	if (ret)
 		return ret;
 
+	/* Track user-specified console flow control. */
+	uart_set_cons_flow_enabled(port, flow == 'r');
+
 	if (port->dev)
 		pm_runtime_get_sync(port->dev);
 
diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
index 89cebdd278410..840336f95c5f6 100644
--- a/drivers/tty/serial/serial_core.c
+++ b/drivers/tty/serial/serial_core.c
@@ -2235,6 +2235,18 @@ uart_set_options(struct uart_port *port, struct console *co,
 	port->mctrl |= TIOCM_DTR;
 
 	port->ops->set_termios(port, &termios, &dummy);
+
+	/*
+	 * If console hardware flow control was specified and is supported,
+	 * the related policy UPSTAT_CTS_ENABLE must be set to allow console
+	 * drivers to identify if CTS should be used for polling.
+	 */
+	if (flow == 'r' && (termios.c_cflag & CRTSCTS)) {
+		/* Synchronize @status RMW update against the console. */
+		guard(uart_port_lock_irqsave)(port);
+		port->status |= UPSTAT_CTS_ENABLE;
+	}
+
 	/*
 	 * Allow the setting of the UART parameters with a NULL console
 	 * too:
@@ -2541,7 +2553,14 @@ uart_configure_port(struct uart_driver *drv, struct uart_state *state,
 		 * We probably don't need a spinlock around this, but
 		 */
 		scoped_guard(uart_port_lock_irqsave, port) {
-			port->mctrl &= TIOCM_DTR;
+			unsigned int mask = TIOCM_DTR;
+
+			/* Console hardware flow control polls CTS. */
+			if (uart_console_hwflow_active(port))
+				mask |= TIOCM_RTS;
+
+			port->mctrl &= mask;
+
 			if (!(port->rs485.flags & SER_RS485_ENABLED))
 				port->ops->set_mctrl(port, port->mctrl);
 		}
diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h
index 4f7bbdd900176..17fcff466e301 100644
--- a/include/linux/serial_core.h
+++ b/include/linux/serial_core.h
@@ -1175,6 +1175,14 @@ static inline bool uart_cons_flow_enabled(const struct uart_port *uport)
 	return uport->cons_flow;
 }
 
+static inline bool uart_console_hwflow_active(struct uart_port *uport)
+{
+	return uart_console(uport) &&
+	       !(uport->rs485.flags & SER_RS485_ENABLED) &&
+	       uart_cons_flow_enabled(uport) &&
+	       uart_cts_enabled(uport);
+}
+
 /*
  * The following are helper functions for the low level drivers.
  */
-- 
2.47.3

Re: [PATCH tty v4 3/6] serial: sh-sci: Avoid deprecated UPF_CONS_FLOW

[Geert Uytterhoeven]

Hi John,

On Wed, 6 May 2026 at 14:16, John Ogness <john.ogness@linutronix.de> wrote:
> Avoid setting the uart_port.flags deprecated UPF_CONS_FLOW bit if it
> has been configured in the platform data. Use the new cons_flow
> wrappers instead.
>
> Signed-off-by: John Ogness <john.ogness@linutronix.de>

Thanks for your patch!

> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -3369,9 +3369,12 @@ static int sci_init_single(struct platform_device *dev,
>         }
>
>         port->type              = SCI_PUBLIC_PORT_ID(p->type);
> -       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF | p->flags;
> +       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF |
> +                                 (p->flags & ~UPF_CONS_FLOW);

This seems over-cautious to me.
The last setter of p->flags was removed in commit 37744feebc086908
("sh: remove sh5 support") in v5.8.  No platform data ever set the
UPF_CONS_FLOW flag before.  I would rather remove plat_sci_port.flags
and this "| p->flags", so we don't have to care about UPF_CONS_FLOW
in this driver at all.

>         port->fifosize          = sci_port->params->fifosize;
>
> +       uart_set_cons_flow_enabled(port, p->flags & UPF_CONS_FLOW);
> +
>         if (p->type == PORT_SCI && !dev->dev.of_node) {
>                 if (sci_port->reg_size >= 0x20)
>                         port->regshift = 2;

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH tty v4 3/6] serial: sh-sci: Avoid deprecated UPF_CONS_FLOW

[Geert Uytterhoeven]

On Wed, 6 May 2026 at 14:39, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> On Wed, 6 May 2026 at 14:16, John Ogness <john.ogness@linutronix.de> wrote:
> > Avoid setting the uart_port.flags deprecated UPF_CONS_FLOW bit if it
> > has been configured in the platform data. Use the new cons_flow
> > wrappers instead.
> >
> > Signed-off-by: John Ogness <john.ogness@linutronix.de>
>
> Thanks for your patch!
>
> > --- a/drivers/tty/serial/sh-sci.c
> > +++ b/drivers/tty/serial/sh-sci.c
> > @@ -3369,9 +3369,12 @@ static int sci_init_single(struct platform_device *dev,
> >         }
> >
> >         port->type              = SCI_PUBLIC_PORT_ID(p->type);
> > -       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF | p->flags;
> > +       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF |
> > +                                 (p->flags & ~UPF_CONS_FLOW);
>
> This seems over-cautious to me.
> The last setter of p->flags was removed in commit 37744feebc086908
> ("sh: remove sh5 support") in v5.8.  No platform data ever set the
> UPF_CONS_FLOW flag before.  I would rather remove plat_sci_port.flags
> and this "| p->flags", so we don't have to care about UPF_CONS_FLOW
> in this driver at all.

I have sent a patch to remove it:
https://lore.kernel.org/20260506124643.128021-1-geert+renesas@glider.be/

Builds on Renesas ARM/ARM64/RISCV/SH, and boots on QEMU
(rts7751)r2d.

>
> >         port->fifosize          = sci_port->params->fifosize;
> >
> > +       uart_set_cons_flow_enabled(port, p->flags & UPF_CONS_FLOW);
> > +
> >         if (p->type == PORT_SCI && !dev->dev.of_node) {
> >                 if (sci_port->reg_size >= 0x20)
> >                         port->regshift = 2;

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH tty v4 3/6] serial: sh-sci: Avoid deprecated UPF_CONS_FLOW

[John Ogness]

Hi Geert,

On 2026-05-06, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>> --- a/drivers/tty/serial/sh-sci.c
>> +++ b/drivers/tty/serial/sh-sci.c
>> @@ -3369,9 +3369,12 @@ static int sci_init_single(struct platform_device *dev,
>>         }
>>
>>         port->type              = SCI_PUBLIC_PORT_ID(p->type);
>> -       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF | p->flags;
>> +       port->flags             = UPF_FIXED_PORT | UPF_BOOT_AUTOCONF |
>> +                                 (p->flags & ~UPF_CONS_FLOW);
>
> This seems over-cautious to me.
> The last setter of p->flags was removed in commit 37744feebc086908
> ("sh: remove sh5 support") in v5.8.  No platform data ever set the
> UPF_CONS_FLOW flag before.  I would rather remove plat_sci_port.flags
> and this "| p->flags", so we don't have to care about UPF_CONS_FLOW
> in this driver at all.

If there is a v5, I will drop this patch. If v4 is acceptable, the
maintainer can just drop this patch.

I will leave the plat_sci_port.flags removal as an excercise for the sh
folks.

Thanks for your feedback.

John