usb_udc_vbus_handler() can be invoked from interrupt context by irq
handlers of the gadget drivers, however, usb_udc_connect_control() has
to run in non-atomic context due to the following:
a. Some of the gadget driver implementations expect the ->pullup
callback to be invoked in non-atomic context.
b. usb_gadget_disconnect() acquires udc_lock which is a mutex.
Hence offload invocation of usb_udc_connect_control()
to workqueue.
Cc: stable(a)vger.kernel.org
Fixes: 1016fc0c096c ("USB: gadget: Fix obscure lockdep violation for udc_mutex")
Signed-off-by: Badhri Jagan Sridharan <badhri(a)google.com>
---
Changes since v1:
- Address Alan Stern's comment on usb_udc_vbus_handler invocation from
atomic context:
* vbus_events_lock is now a spinlock and allocations in
* usb_udc_vbus_handler are atomic now.
Changes since v2:
- Addressing Alan Stern's comments:
** connect_lock is now held by callers of
* usb_gadget_pullup_update_locked() and gadget_(un)bind_driver() does
* notdirectly hold the lock.
** Both usb_gadget_(dis)connect() and usb_udc_vbus_handler() would
* set/clear udc->vbus and invoke usb_gadget_pullup_update_locked.
** Add "unbinding" to prevent new connections after the gadget is being
* unbound.
Changes since v3:
** Made a minor cleanup which I missed to do in v3 in
* usb_udc_vbus_handler().
Changes since v4:
- Addressing Alan Stern's comments:
** usb_udc_vbus_handler() now offloads invocation of usb_udc_connect_control()
* from workqueue.
** Dropped vbus_events list as this was redundant. Updating to the
* latest value is suffice
---
drivers/usb/gadget/udc/core.c | 19 ++++++++++++++++++-
1 file changed, 18 insertions(+), 1 deletion(-)
diff --git a/drivers/usb/gadget/udc/core.c b/drivers/usb/gadget/udc/core.c
index 52e6d2e84e35..44a9f32679b5 100644
--- a/drivers/usb/gadget/udc/core.c
+++ b/drivers/usb/gadget/udc/core.c
@@ -48,6 +48,7 @@ struct usb_udc {
struct list_head list;
bool vbus;
bool started;
+ struct work_struct vbus_work;
};
static struct class *udc_class;
@@ -1086,6 +1087,13 @@ static void usb_udc_connect_control(struct usb_udc *udc)
usb_gadget_disconnect(udc->gadget);
}
+static void vbus_event_work(struct work_struct *work)
+{
+ struct usb_udc *udc = container_of(work, struct usb_udc, vbus_work);
+
+ usb_udc_connect_control(udc);
+}
+
/**
* usb_udc_vbus_handler - updates the udc core vbus status, and try to
* connect or disconnect gadget
@@ -1094,6 +1102,13 @@ static void usb_udc_connect_control(struct usb_udc *udc)
*
* The udc driver calls it when it wants to connect or disconnect gadget
* according to vbus status.
+ *
+ * This function can be invoked from interrupt context by irq handlers of the gadget drivers,
+ * however, usb_udc_connect_control() has to run in non-atomic context due to the following:
+ * a. Some of the gadget driver implementations expect the ->pullup callback to be invoked in
+ * non-atomic context.
+ * b. usb_gadget_disconnect() acquires udc_lock which is a mutex.
+ * Hence offload invocation of usb_udc_connect_control() to workqueue.
*/
void usb_udc_vbus_handler(struct usb_gadget *gadget, bool status)
{
@@ -1101,7 +1116,7 @@ void usb_udc_vbus_handler(struct usb_gadget *gadget, bool status)
if (udc) {
udc->vbus = status;
- usb_udc_connect_control(udc);
+ schedule_work(&udc->vbus_work);
}
}
EXPORT_SYMBOL_GPL(usb_udc_vbus_handler);
@@ -1328,6 +1343,7 @@ int usb_add_gadget(struct usb_gadget *gadget)
mutex_lock(&udc_lock);
list_add_tail(&udc->list, &udc_list);
mutex_unlock(&udc_lock);
+ INIT_WORK(&udc->vbus_work, vbus_event_work);
ret = device_add(&udc->dev);
if (ret)
@@ -1558,6 +1574,7 @@ static void gadget_unbind_driver(struct device *dev)
kobject_uevent(&udc->dev.kobj, KOBJ_CHANGE);
+ cancel_work_sync(&udc->vbus_work);
usb_gadget_disconnect(gadget);
usb_gadget_disable_async_callbacks(udc);
if (gadget->irq)
base-commit: 046895105d9666ab56e86ce8dd9786f8003125c6
--
2.41.0.rc0.172.g3f132b7071-goog
From: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
When configuring a pin as an output pin with a value of logic 0, we
end up as having a value of logic 1 on the output pin. Setting a
logic 0 a second time (or more) after that will correctly output a
logic 0 on the output pin.
By default, all GPIO pins are configured as inputs. When we enter
sc16is7xx_gpio_direction_output() for the first time, we first set the
desired value in IOSTATE, and then we configure the pin as an output.
The datasheet states that writing to IOSTATE register will trigger a
transfer of the value to the I/O pin configured as output, so if the
pin is configured as an input, nothing will be transferred.
Therefore, set the direction first in IODIR, and then set the desired
value in IOSTATE.
This is what is done in NXP application note AN10587.
Fixes: dfeae619d781 ("serial: sc16is7xx")
Cc: <stable(a)vger.kernel.org>
Signed-off-by: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Reviewed-by: Lech Perczak <lech.perczak(a)camlingroup.com>
Tested-by: Lech Perczak <lech.perczak(a)camlingroup.com>
---
drivers/tty/serial/sc16is7xx.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/serial/sc16is7xx.c b/drivers/tty/serial/sc16is7xx.c
index edc83f5f6340..7a9b91f0c710 100644
--- a/drivers/tty/serial/sc16is7xx.c
+++ b/drivers/tty/serial/sc16is7xx.c
@@ -1340,9 +1340,18 @@ static int sc16is7xx_gpio_direction_output(struct gpio_chip *chip,
state |= BIT(offset);
else
state &= ~BIT(offset);
- sc16is7xx_port_write(port, SC16IS7XX_IOSTATE_REG, state);
+
+ /*
+ * If we write IOSTATE first, and then IODIR, the output value is not
+ * transferred to the corresponding I/O pin.
+ * The datasheet states that each register bit will be transferred to
+ * the corresponding I/O pin programmed as output when writing to
+ * IOSTATE. Therefore, configure direction first with IODIR, and then
+ * set value after with IOSTATE.
+ */
sc16is7xx_port_update(port, SC16IS7XX_IODIR_REG, BIT(offset),
BIT(offset));
+ sc16is7xx_port_write(port, SC16IS7XX_IOSTATE_REG, state);
return 0;
}
--
2.30.2
From: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Some variants in this series of UART controllers have GPIO pins that
are shared between GPIO and modem control lines.
The pin mux mode (GPIO or modem control lines) can be set for each
ports (channels) supported by the variant.
This adds a property to the device tree to set the GPIO pin mux to
modem control lines on selected ports if needed.
Cc: <stable(a)vger.kernel.org> # 6.1.x
Signed-off-by: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Acked-by: Conor Dooley <conor.dooley(a)microchip.com>
Reviewed-by: Lech Perczak <lech.perczak(a)camlingroup.com>
---
.../bindings/serial/nxp,sc16is7xx.txt | 46 +++++++++++++++++++
1 file changed, 46 insertions(+)
diff --git a/Documentation/devicetree/bindings/serial/nxp,sc16is7xx.txt b/Documentation/devicetree/bindings/serial/nxp,sc16is7xx.txt
index 0fa8e3e43bf8..1a7e4bff0456 100644
--- a/Documentation/devicetree/bindings/serial/nxp,sc16is7xx.txt
+++ b/Documentation/devicetree/bindings/serial/nxp,sc16is7xx.txt
@@ -23,6 +23,9 @@ Optional properties:
1 = active low.
- irda-mode-ports: An array that lists the indices of the port that
should operate in IrDA mode.
+- nxp,modem-control-line-ports: An array that lists the indices of the port that
+ should have shared GPIO lines configured as
+ modem control lines.
Example:
sc16is750: sc16is750@51 {
@@ -35,6 +38,26 @@ Example:
#gpio-cells = <2>;
};
+ sc16is752: sc16is752@53 {
+ compatible = "nxp,sc16is752";
+ reg = <0x53>;
+ clocks = <&clk20m>;
+ interrupt-parent = <&gpio3>;
+ interrupts = <7 IRQ_TYPE_EDGE_FALLING>;
+ nxp,modem-control-line-ports = <1>; /* Port 1 as modem control lines */
+ gpio-controller; /* Port 0 as GPIOs */
+ #gpio-cells = <2>;
+ };
+
+ sc16is752: sc16is752@54 {
+ compatible = "nxp,sc16is752";
+ reg = <0x54>;
+ clocks = <&clk20m>;
+ interrupt-parent = <&gpio3>;
+ interrupts = <7 IRQ_TYPE_EDGE_FALLING>;
+ nxp,modem-control-line-ports = <0 1>; /* Ports 0 and 1 as modem control lines */
+ };
+
* spi as bus
Required properties:
@@ -59,6 +82,9 @@ Optional properties:
1 = active low.
- irda-mode-ports: An array that lists the indices of the port that
should operate in IrDA mode.
+- nxp,modem-control-line-ports: An array that lists the indices of the port that
+ should have shared GPIO lines configured as
+ modem control lines.
Example:
sc16is750: sc16is750@0 {
@@ -70,3 +96,23 @@ Example:
gpio-controller;
#gpio-cells = <2>;
};
+
+ sc16is752: sc16is752@1 {
+ compatible = "nxp,sc16is752";
+ reg = <1>;
+ clocks = <&clk20m>;
+ interrupt-parent = <&gpio3>;
+ interrupts = <7 IRQ_TYPE_EDGE_FALLING>;
+ nxp,modem-control-line-ports = <1>; /* Port 1 as modem control lines */
+ gpio-controller; /* Port 0 as GPIOs */
+ #gpio-cells = <2>;
+ };
+
+ sc16is752: sc16is752@2 {
+ compatible = "nxp,sc16is752";
+ reg = <2>;
+ clocks = <&clk20m>;
+ interrupt-parent = <&gpio3>;
+ interrupts = <7 IRQ_TYPE_EDGE_FALLING>;
+ nxp,modem-control-line-ports = <0 1>; /* Ports 0 and 1 as modem control lines */
+ };
--
2.30.2
From: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Bit SRESET (3) is cleared when a reset operation is completed. Having
the IOCONTROL register as non-volatile will always read SRESET as 1,
which is incorrect.
Also, if IOCONTROL register is not a volatile register, the upcoming
patch "serial: sc16is7xx: fix regression with GPIO configuration"
doesn't work when setting some shared GPIO lines as modem control
lines.
Therefore mark IOCONTROL register as a volatile register.
Cc: <stable(a)vger.kernel.org> # 6.1.x
Signed-off-by: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Reviewed-by: Lech Perczak <lech.perczak(a)camlingroup.com>
Tested-by: Lech Perczak <lech.perczak(a)camlingroup.com>
---
drivers/tty/serial/sc16is7xx.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/tty/serial/sc16is7xx.c b/drivers/tty/serial/sc16is7xx.c
index faa51a58671f..0c903d44429c 100644
--- a/drivers/tty/serial/sc16is7xx.c
+++ b/drivers/tty/serial/sc16is7xx.c
@@ -488,6 +488,7 @@ static bool sc16is7xx_regmap_volatile(struct device *dev, unsigned int reg)
case SC16IS7XX_TXLVL_REG:
case SC16IS7XX_RXLVL_REG:
case SC16IS7XX_IOSTATE_REG:
+ case SC16IS7XX_IOCONTROL_REG:
return true;
default:
break;
--
2.30.2
From: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
The sc16is7xx_config_rs485() function is called only for the second
port (index 1, channel B), causing initialization problems for the
first port.
For the sc16is7xx driver, port->membase and port->mapbase are not set,
and their default values are 0. And we set port->iobase to the device
index. This means that when the first device is registered using the
uart_add_one_port() function, the following values will be in the port
structure:
port->membase = 0
port->mapbase = 0
port->iobase = 0
Therefore, the function uart_configure_port() in serial_core.c will
exit early because of the following check:
/*
* If there isn't a port here, don't do anything further.
*/
if (!port->iobase && !port->mapbase && !port->membase)
return;
Typically, I2C and SPI drivers do not set port->membase and
port->mapbase.
The max310x driver sets port->membase to ~0 (all ones). By
implementing the same change in this driver, uart_configure_port() is
now correctly executed for all ports.
Fixes: dfeae619d781 ("serial: sc16is7xx")
Cc: <stable(a)vger.kernel.org> # 6.1.x
Signed-off-by: Hugo Villeneuve <hvilleneuve(a)dimonoff.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen(a)linux.intel.com>
Reviewed-by: Lech Perczak <lech.perczak(a)camlingroup.com>
Tested-by: Lech Perczak <lech.perczak(a)camlingroup.com>
---
drivers/tty/serial/sc16is7xx.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/tty/serial/sc16is7xx.c b/drivers/tty/serial/sc16is7xx.c
index abad091baeea..faa51a58671f 100644
--- a/drivers/tty/serial/sc16is7xx.c
+++ b/drivers/tty/serial/sc16is7xx.c
@@ -1436,6 +1436,7 @@ static int sc16is7xx_probe(struct device *dev,
s->p[i].port.fifosize = SC16IS7XX_FIFO_SIZE;
s->p[i].port.flags = UPF_FIXED_TYPE | UPF_LOW_LATENCY;
s->p[i].port.iobase = i;
+ s->p[i].port.membase = (void __iomem *)~0;
s->p[i].port.iotype = UPIO_PORT;
s->p[i].port.uartclk = freq;
s->p[i].port.rs485_config = sc16is7xx_config_rs485;
--
2.30.2