2014-10-10 13:38:18

by Ricardo Ribalda Delgado

[permalink] [raw]
Subject: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

Introduce an homogeneous lock system between setting and using the rs485
data of the uart_port.

This patch should not be split into multiple ones in order to avoid
leaving the tree in an unstable state.

Suggested-by: Alan Cox <[email protected]>
Cc: Nicolas Ferre <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Jiri Slaby <[email protected]>
Cc: One Thousand Gnomes <[email protected]>
Signed-off-by: Ricardo Ribalda Delgado <[email protected]>
---


This patch sits on top of the series: "[PATCH 00/12] Handle TIOC[GS]RS485
iocts on serial_core"

If there are major changes on the other patches I will resend the whole
patchset with this patch as part of it

drivers/tty/serial/atmel_serial.c | 10 ++++------
drivers/tty/serial/mcf.c | 5 +----
drivers/tty/serial/omap-serial.c | 3 ---
drivers/tty/serial/serial_core.c | 13 ++++++++++++-
4 files changed, 17 insertions(+), 14 deletions(-)

diff --git a/drivers/tty/serial/atmel_serial.c b/drivers/tty/serial/atmel_serial.c
index e8df8e4..32b2a8d 100644
--- a/drivers/tty/serial/atmel_serial.c
+++ b/drivers/tty/serial/atmel_serial.c
@@ -293,9 +293,6 @@ static int atmel_config_rs485(struct uart_port *port,
{
struct atmel_uart_port *atmel_port = to_atmel_uart_port(port);
unsigned int mode;
- unsigned long flags;
-
- spin_lock_irqsave(&port->lock, flags);

/* Disable interrupts */
UART_PUT_IDR(port, atmel_port->tx_done_mask);
@@ -326,8 +323,6 @@ static int atmel_config_rs485(struct uart_port *port,
/* Enable interrupts */
UART_PUT_IER(port, atmel_port->tx_done_mask);

- spin_unlock_irqrestore(&port->lock, flags);
-
return 0;
}

@@ -2509,6 +2504,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
struct atmel_uart_data *pdata = dev_get_platdata(&pdev->dev);
void *data;
int ret = -ENODEV;
+ bool rs485_enabled;

BUILD_BUG_ON(ATMEL_SERIAL_RINGSIZE & (ATMEL_SERIAL_RINGSIZE - 1));

@@ -2556,6 +2552,8 @@ static int atmel_serial_probe(struct platform_device *pdev)
port->rx_ring.buf = data;
}

+ rs485_enabled = port->uart.rs485.flags & SER_RS485_ENABLED;
+
ret = uart_add_one_port(&atmel_uart, &port->uart);
if (ret)
goto err_add_port;
@@ -2574,7 +2572,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
device_init_wakeup(&pdev->dev, 1);
platform_set_drvdata(pdev, port);

- if (port->uart.rs485.flags & SER_RS485_ENABLED) {
+ if (rs485_enabled) {
UART_PUT_MR(&port->uart, ATMEL_US_USMODE_NORMAL);
UART_PUT_CR(&port->uart, ATMEL_US_RTSEN);
}
diff --git a/drivers/tty/serial/mcf.c b/drivers/tty/serial/mcf.c
index d7be1f1..fdd5c7b 100644
--- a/drivers/tty/serial/mcf.c
+++ b/drivers/tty/serial/mcf.c
@@ -257,12 +257,12 @@ static void mcf_set_termios(struct uart_port *port, struct ktermios *termios,
mr2 |= MCFUART_MR2_TXCTS;
}

+ spin_lock_irqsave(&port->lock, flags);
if (port->rs485.flags & SER_RS485_ENABLED) {
dev_dbg(port->dev, "Setting UART to RS485\n");
mr2 |= MCFUART_MR2_TXRTS;
}

- spin_lock_irqsave(&port->lock, flags);
uart_update_timeout(port, termios->c_cflag, baud);
writeb(MCFUART_UCR_CMDRESETRX, port->membase + MCFUART_UCR);
writeb(MCFUART_UCR_CMDRESETTX, port->membase + MCFUART_UCR);
@@ -442,10 +442,8 @@ static int mcf_verify_port(struct uart_port *port, struct serial_struct *ser)
static int mcf_config_rs485(struct uart_port *port, struct serial_rs485 *rs485)
{
struct mcf_uart *pp = container_of(port, struct mcf_uart, port);
- unsigned long flags;
unsigned char mr1, mr2;

- spin_lock_irqsave(&port->lock, flags);
/* Get mode registers */
mr1 = readb(port->membase + MCFUART_UMR);
mr2 = readb(port->membase + MCFUART_UMR);
@@ -460,7 +458,6 @@ static int mcf_config_rs485(struct uart_port *port, struct serial_rs485 *rs485)
writeb(mr1, port->membase + MCFUART_UMR);
writeb(mr2, port->membase + MCFUART_UMR);
port->rs485 = *rs485;
- spin_unlock_irqrestore(&port->lock, flags);

return 0;
}
diff --git a/drivers/tty/serial/omap-serial.c b/drivers/tty/serial/omap-serial.c
index 9d4726f..6675dbf 100644
--- a/drivers/tty/serial/omap-serial.c
+++ b/drivers/tty/serial/omap-serial.c
@@ -1360,12 +1360,10 @@ static int
serial_omap_config_rs485(struct uart_port *port, struct serial_rs485 *rs485conf)
{
struct uart_omap_port *up = to_uart_omap_port(port);
- unsigned long flags;
unsigned int mode;
int val;

pm_runtime_get_sync(up->dev);
- spin_lock_irqsave(&up->port.lock, flags);

/* Disable interrupts from this port */
mode = up->ier;
@@ -1401,7 +1399,6 @@ serial_omap_config_rs485(struct uart_port *port, struct serial_rs485 *rs485conf)
serial_out(up, UART_OMAP_SCR, up->scr);
}

- spin_unlock_irqrestore(&up->port.lock, flags);
pm_runtime_mark_last_busy(up->dev);
pm_runtime_put_autosuspend(up->dev);

diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
index b59c925..b574dcd 100644
--- a/drivers/tty/serial/serial_core.c
+++ b/drivers/tty/serial/serial_core.c
@@ -1154,8 +1154,16 @@ static int uart_get_icount(struct tty_struct *tty,
static int uart_get_rs485_config(struct uart_port *port,
struct serial_rs485 __user *rs485)
{
- if (copy_to_user(rs485, &port->rs485, sizeof(port->rs485)))
+ unsigned long flags;
+ int ret;
+
+ spin_lock_irqsave(&port->lock, flags);
+ ret = copy_to_user(rs485, &port->rs485, sizeof(port->rs485));
+ spin_unlock_irqrestore(&port->lock, flags);
+
+ if (ret)
return -EFAULT;
+
return 0;
}

@@ -1164,6 +1172,7 @@ static int uart_set_rs485_config(struct uart_port *port,
{
struct serial_rs485 rs485;
int ret;
+ unsigned long flags;

if (!port->rs485_config)
return -ENOIOCTLCMD;
@@ -1171,7 +1180,9 @@ static int uart_set_rs485_config(struct uart_port *port,
if (copy_from_user(&rs485, rs485_user, sizeof(rs485_user)))
return -EFAULT;

+ spin_lock_irqsave(&port->lock, flags);
ret = port->rs485_config(port, &rs485);
+ spin_unlock_irqrestore(&port->lock, flags);
if (ret)
return ret;

--
2.1.1


2014-10-12 05:23:43

by Nicolas Ferre

[permalink] [raw]
Subject: Re: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

On 10/10/2014 15:38, Ricardo Ribalda Delgado :
> Introduce an homogeneous lock system between setting and using the rs485
> data of the uart_port.
>
> This patch should not be split into multiple ones in order to avoid
> leaving the tree in an unstable state.
>
> Suggested-by: Alan Cox <[email protected]>
> Cc: Nicolas Ferre <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: Jiri Slaby <[email protected]>
> Cc: One Thousand Gnomes <[email protected]>
> Signed-off-by: Ricardo Ribalda Delgado <[email protected]>
> ---
>
>
> This patch sits on top of the series: "[PATCH 00/12] Handle TIOC[GS]RS485
> iocts on serial_core"
>
> If there are major changes on the other patches I will resend the whole
> patchset with this patch as part of it
>
> drivers/tty/serial/atmel_serial.c | 10 ++++------
> drivers/tty/serial/mcf.c | 5 +----
> drivers/tty/serial/omap-serial.c | 3 ---
> drivers/tty/serial/serial_core.c | 13 ++++++++++++-
> 4 files changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/tty/serial/atmel_serial.c b/drivers/tty/serial/atmel_serial.c
> index e8df8e4..32b2a8d 100644
> --- a/drivers/tty/serial/atmel_serial.c
> +++ b/drivers/tty/serial/atmel_serial.c
> @@ -293,9 +293,6 @@ static int atmel_config_rs485(struct uart_port *port,
> {
> struct atmel_uart_port *atmel_port = to_atmel_uart_port(port);
> unsigned int mode;
> - unsigned long flags;
> -
> - spin_lock_irqsave(&port->lock, flags);
>
> /* Disable interrupts */
> UART_PUT_IDR(port, atmel_port->tx_done_mask);
> @@ -326,8 +323,6 @@ static int atmel_config_rs485(struct uart_port *port,
> /* Enable interrupts */
> UART_PUT_IER(port, atmel_port->tx_done_mask);
>
> - spin_unlock_irqrestore(&port->lock, flags);
> -
> return 0;
> }
>
> @@ -2509,6 +2504,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
> struct atmel_uart_data *pdata = dev_get_platdata(&pdev->dev);
> void *data;
> int ret = -ENODEV;
> + bool rs485_enabled;
>
> BUILD_BUG_ON(ATMEL_SERIAL_RINGSIZE & (ATMEL_SERIAL_RINGSIZE - 1));
>
> @@ -2556,6 +2552,8 @@ static int atmel_serial_probe(struct platform_device *pdev)
> port->rx_ring.buf = data;
> }
>
> + rs485_enabled = port->uart.rs485.flags & SER_RS485_ENABLED;
> +

I have the feeling that moving the code chunk that uses this new
variable (rs485_enabled) here ...

> ret = uart_add_one_port(&atmel_uart, &port->uart);
> if (ret)
> goto err_add_port;
> @@ -2574,7 +2572,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
> device_init_wakeup(&pdev->dev, 1);
> platform_set_drvdata(pdev, port);
>
> - if (port->uart.rs485.flags & SER_RS485_ENABLED) {
> + if (rs485_enabled) {
> UART_PUT_MR(&port->uart, ATMEL_US_USMODE_NORMAL);
> UART_PUT_CR(&port->uart, ATMEL_US_RTSEN);
> }

... (this one ^^^) up where you can test the SER_RS485_ENABLED, can be
even simpler.

otherwise, it seems good so you can add my:

Acked-by: Nicolas Ferre <[email protected]>

Thanks.

> diff --git a/drivers/tty/serial/mcf.c b/drivers/tty/serial/mcf.c
> index d7be1f1..fdd5c7b 100644
> --- a/drivers/tty/serial/mcf.c
> +++ b/drivers/tty/serial/mcf.c
> @@ -257,12 +257,12 @@ static void mcf_set_termios(struct uart_port *port, struct ktermios *termios,
> mr2 |= MCFUART_MR2_TXCTS;
> }
>
> + spin_lock_irqsave(&port->lock, flags);
> if (port->rs485.flags & SER_RS485_ENABLED) {
> dev_dbg(port->dev, "Setting UART to RS485\n");
> mr2 |= MCFUART_MR2_TXRTS;
> }
>
> - spin_lock_irqsave(&port->lock, flags);
> uart_update_timeout(port, termios->c_cflag, baud);
> writeb(MCFUART_UCR_CMDRESETRX, port->membase + MCFUART_UCR);
> writeb(MCFUART_UCR_CMDRESETTX, port->membase + MCFUART_UCR);
> @@ -442,10 +442,8 @@ static int mcf_verify_port(struct uart_port *port, struct serial_struct *ser)
> static int mcf_config_rs485(struct uart_port *port, struct serial_rs485 *rs485)
> {
> struct mcf_uart *pp = container_of(port, struct mcf_uart, port);
> - unsigned long flags;
> unsigned char mr1, mr2;
>
> - spin_lock_irqsave(&port->lock, flags);
> /* Get mode registers */
> mr1 = readb(port->membase + MCFUART_UMR);
> mr2 = readb(port->membase + MCFUART_UMR);
> @@ -460,7 +458,6 @@ static int mcf_config_rs485(struct uart_port *port, struct serial_rs485 *rs485)
> writeb(mr1, port->membase + MCFUART_UMR);
> writeb(mr2, port->membase + MCFUART_UMR);
> port->rs485 = *rs485;
> - spin_unlock_irqrestore(&port->lock, flags);
>
> return 0;
> }
> diff --git a/drivers/tty/serial/omap-serial.c b/drivers/tty/serial/omap-serial.c
> index 9d4726f..6675dbf 100644
> --- a/drivers/tty/serial/omap-serial.c
> +++ b/drivers/tty/serial/omap-serial.c
> @@ -1360,12 +1360,10 @@ static int
> serial_omap_config_rs485(struct uart_port *port, struct serial_rs485 *rs485conf)
> {
> struct uart_omap_port *up = to_uart_omap_port(port);
> - unsigned long flags;
> unsigned int mode;
> int val;
>
> pm_runtime_get_sync(up->dev);
> - spin_lock_irqsave(&up->port.lock, flags);
>
> /* Disable interrupts from this port */
> mode = up->ier;
> @@ -1401,7 +1399,6 @@ serial_omap_config_rs485(struct uart_port *port, struct serial_rs485 *rs485conf)
> serial_out(up, UART_OMAP_SCR, up->scr);
> }
>
> - spin_unlock_irqrestore(&up->port.lock, flags);
> pm_runtime_mark_last_busy(up->dev);
> pm_runtime_put_autosuspend(up->dev);
>
> diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
> index b59c925..b574dcd 100644
> --- a/drivers/tty/serial/serial_core.c
> +++ b/drivers/tty/serial/serial_core.c
> @@ -1154,8 +1154,16 @@ static int uart_get_icount(struct tty_struct *tty,
> static int uart_get_rs485_config(struct uart_port *port,
> struct serial_rs485 __user *rs485)
> {
> - if (copy_to_user(rs485, &port->rs485, sizeof(port->rs485)))
> + unsigned long flags;
> + int ret;
> +
> + spin_lock_irqsave(&port->lock, flags);
> + ret = copy_to_user(rs485, &port->rs485, sizeof(port->rs485));
> + spin_unlock_irqrestore(&port->lock, flags);
> +
> + if (ret)
> return -EFAULT;
> +
> return 0;
> }
>
> @@ -1164,6 +1172,7 @@ static int uart_set_rs485_config(struct uart_port *port,
> {
> struct serial_rs485 rs485;
> int ret;
> + unsigned long flags;
>
> if (!port->rs485_config)
> return -ENOIOCTLCMD;
> @@ -1171,7 +1180,9 @@ static int uart_set_rs485_config(struct uart_port *port,
> if (copy_from_user(&rs485, rs485_user, sizeof(rs485_user)))
> return -EFAULT;
>
> + spin_lock_irqsave(&port->lock, flags);
> ret = port->rs485_config(port, &rs485);
> + spin_unlock_irqrestore(&port->lock, flags);
> if (ret)
> return ret;
>
>


--
Nicolas Ferre

2014-10-13 13:52:48

by Alan Cox

[permalink] [raw]
Subject: Re: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

> @@ -1154,8 +1154,16 @@ static int uart_get_icount(struct tty_struct *tty,
> static int uart_get_rs485_config(struct uart_port *port,
> struct serial_rs485 __user *rs485)
> {
> - if (copy_to_user(rs485, &port->rs485, sizeof(port->rs485)))
> + unsigned long flags;
> + int ret;
> +
> + spin_lock_irqsave(&port->lock, flags);
> + ret = copy_to_user(rs485, &port->rs485, sizeof(port->rs485));
> + spin_unlock_irqrestore(&port->lock, flags);

Looks good apart from this bit.. The copy needs to go to a temporary
before the lock is taken otherwise we may sleep on a pagefault with irqs
off

Alan

2014-10-13 19:03:22

by Ricardo Ribalda Delgado

[permalink] [raw]
Subject: Re: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

Hello Alan

>> + spin_unlock_irqrestore(&port->lock, flags);
>
> Looks good apart from this bit.. The copy needs to go to a temporary
> before the lock is taken otherwise we may sleep on a pagefault with irqs
> off
>
> Alan

Good catch

I will fix it as you suggest.

I will give a give a couple of days for comments about the serie
"[PATCH 00/12] Handle TIOC[GS]RS485
iocts on serial_core" Then I will resend the series including this
patch (of course, fixed)


Thanks!


--
Ricardo Ribalda

2014-10-13 19:03:42

by Ricardo Ribalda Delgado

[permalink] [raw]
Subject: Re: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

Hello Nicolas

>
> I have the feeling that moving the code chunk that uses this new
> variable (rs485_enabled) here ...
>
>> ret = uart_add_one_port(&atmel_uart, &port->uart);
>> if (ret)
>> goto err_add_port;
>> @@ -2574,7 +2572,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
>> device_init_wakeup(&pdev->dev, 1);
>> platform_set_drvdata(pdev, port);
>>
>> - if (port->uart.rs485.flags & SER_RS485_ENABLED) {
>> + if (rs485_enabled) {
>> UART_PUT_MR(&port->uart, ATMEL_US_USMODE_NORMAL);
>> UART_PUT_CR(&port->uart, ATMEL_US_RTSEN);
>> }
>
> ... (this one ^^^) up where you can test the SER_RS485_ENABLED, can be
> even simpler.
>
> otherwise, it seems good so you can add my:
>
> Acked-by: Nicolas Ferre <[email protected]>
>
> Thanks.

I supposed that the code needs to be initialized with rs485 disabled
and then enable it. Also I did not want to change the behaviour a
driver in a patch that involves 3 other drivers. If you don't mind I
prefer that this is added to a separated patch.

Thanks

2014-10-13 21:48:13

by Nicolas Ferre

[permalink] [raw]
Subject: Re: [PATCH] tty/serial_core: Introduce lock mechanism for RS485

On 13/10/2014 21:03, Ricardo Ribalda Delgado :
> Hello Nicolas
>
>>
>> I have the feeling that moving the code chunk that uses this new
>> variable (rs485_enabled) here ...
>>
>>> ret = uart_add_one_port(&atmel_uart, &port->uart);
>>> if (ret)
>>> goto err_add_port;
>>> @@ -2574,7 +2572,7 @@ static int atmel_serial_probe(struct platform_device *pdev)
>>> device_init_wakeup(&pdev->dev, 1);
>>> platform_set_drvdata(pdev, port);
>>>
>>> - if (port->uart.rs485.flags & SER_RS485_ENABLED) {
>>> + if (rs485_enabled) {
>>> UART_PUT_MR(&port->uart, ATMEL_US_USMODE_NORMAL);
>>> UART_PUT_CR(&port->uart, ATMEL_US_RTSEN);
>>> }
>>
>> ... (this one ^^^) up where you can test the SER_RS485_ENABLED, can be
>> even simpler.
>>
>> otherwise, it seems good so you can add my:
>>
>> Acked-by: Nicolas Ferre <[email protected]>
>>
>> Thanks.
>
> I supposed that the code needs to be initialized with rs485 disabled
> and then enable it. Also I did not want to change the behaviour a
> driver in a patch that involves 3 other drivers. If you don't mind I
> prefer that this is added to a separated patch.

Fair enough, I can do it later on.

Thanks, bye,
--
Nicolas Ferre