Hi,
This patchset fixes a bug in the Synopsis 8250 serial driver which causes the
driver to hang. The bug occurs on simple 2 wire serial ports when modem control
signalling has been enabled. It can be reproduced from user space by enabling
modem control signals (stty -clocal), then opening a serial port and polling for
data available to read with a timeout. A properly implemented driver will
ignore the control signals and the call to poll will return. The current
version of the driver hangs forever on the call to poll despite the timeout.
Jon
Desmond Liu (1):
serial: 8250_dw: Fix get_mctrl behaviour
.../bindings/serial/snps-dw-apb-uart.txt | 7 ++++
drivers/tty/serial/8250/8250_dw.c | 41 ++++++++++++++++++++
2 files changed, 48 insertions(+)
--
1.7.9.5
Fixed behaviour of get_mctrl() serial driver function as documented in:
https://www.kernel.org/doc/Documentation/serial/driver
Added device-tree property 'msr-override' specific to the Synopsis 8250
DesignWare UART driver. Allows one to force Data Carrier Detect,
Clear To Send, and Data Set Ready signals to permanently be reported as
active. The Ring indicator can be forced to be reported as inactive.
It is possible that if modem control signalling is enabled on a port
that doesn't have these pins (e.g. - a simple two wire Tx/Rx port), the
driver can hang indefinitely waiting for the state to change. The new
DT properties allow the driver to ignore the state of these pins on
serial ports that don't support them, as recommended in the kernel
documentation.
Reviewed-by: JD (Jiandong) Zheng <[email protected]>
Reviewed-by: Jonathan Richardson <[email protected]>
Reviewed-by: Scott Branden <[email protected]>
Tested-by: Scott Branden <[email protected]>
Tested-by: Jonathan Richardson <[email protected]>
Signed-off-by: Jonathan Richardson <[email protected]>
---
.../bindings/serial/snps-dw-apb-uart.txt | 7 ++++
drivers/tty/serial/8250/8250_dw.c | 41 ++++++++++++++++++++
2 files changed, 48 insertions(+)
diff --git a/Documentation/devicetree/bindings/serial/snps-dw-apb-uart.txt b/Documentation/devicetree/bindings/serial/snps-dw-apb-uart.txt
index 7f76214..010e52b 100644
--- a/Documentation/devicetree/bindings/serial/snps-dw-apb-uart.txt
+++ b/Documentation/devicetree/bindings/serial/snps-dw-apb-uart.txt
@@ -21,6 +21,12 @@ Optional properties:
- reg-io-width : the size (in bytes) of the IO accesses that should be
performed on the device. If this property is not present then single byte
accesses are used.
+- msr-override : array of strings to be used to override the individual
+ modem status signals for DCD, DSR, CTS, and RI. If the property is not
+ present, the individual signals are obtained from the modem status register.
+ Strings accepted are "dcd", "dsr", "cts", and "ri". If "dcd", "dsr", or
+ "cts" are present, these signals will always be reported as active. If
+ "ri" is present, this signal will always be reported as inactive.
Example:
@@ -31,6 +37,7 @@ Example:
interrupts = <10>;
reg-shift = <2>;
reg-io-width = <4>;
+ msr-override = "dcd", "dsr", cts", "ri"
};
Example with one clock:
diff --git a/drivers/tty/serial/8250/8250_dw.c b/drivers/tty/serial/8250/8250_dw.c
index 555de07..06dc873 100644
--- a/drivers/tty/serial/8250/8250_dw.c
+++ b/drivers/tty/serial/8250/8250_dw.c
@@ -59,6 +59,8 @@ struct dw8250_data {
u8 usr_reg;
int last_mcr;
int line;
+ int msr_mask_on;
+ int msr_mask_off;
struct clk *clk;
struct clk *pclk;
struct reset_control *rst;
@@ -81,6 +83,12 @@ static inline int dw8250_modify_msr(struct uart_port *p, int offset, int value)
value &= ~UART_MSR_DCTS;
}
+ /* Override any modem control signals if needed */
+ if (offset == UART_MSR) {
+ value |= d->msr_mask_on;
+ value &= ~d->msr_mask_off;
+ }
+
return value;
}
@@ -290,6 +298,8 @@ static int dw8250_probe_of(struct uart_port *p,
u32 val;
bool has_ucv = true;
int id;
+ int msr_cnt, i;
+ const char *inp_name;
#ifdef CONFIG_64BIT
if (of_device_is_compatible(np, "cavium,octeon-3860-uart")) {
@@ -334,6 +344,37 @@ static int dw8250_probe_of(struct uart_port *p,
if (id >= 0)
p->line = id;
+ msr_cnt = of_property_count_strings(np, "msr-override");
+
+ if (msr_cnt > 0) {
+ for (i = 0; i < msr_cnt; i++) {
+ of_property_read_string_index(np, "msr-override", i,
+ &inp_name);
+
+ if (!strcmp("dcd", inp_name)) {
+ /* Always report DCD as active */
+ data->msr_mask_on |= UART_MSR_DCD;
+ data->msr_mask_off |= UART_MSR_DDCD;
+ } else if (!strcmp("dsr", inp_name)) {
+ /* Always report DSR as active */
+ data->msr_mask_on |= UART_MSR_DSR;
+ data->msr_mask_off |= UART_MSR_DDSR;
+ } else if (!strcmp("cts", inp_name)) {
+ /* Always report CTS as active */
+ data->msr_mask_on |= UART_MSR_CTS;
+ data->msr_mask_off |= UART_MSR_DCTS;
+ } else if (!strcmp("ri", inp_name)) {
+ /* Always report Ring indicator as inactive */
+ data->msr_mask_off |= UART_MSR_RI;
+ data->msr_mask_off |= UART_MSR_TERI;
+ } else {
+ dev_err(p->dev,
+ "Ignore unknown msr-override %s\n",
+ inp_name);
+ }
+ }
+ }
+
/* clock got configured through clk api, all done */
if (p->uartclk)
return 0;
--
1.7.9.5
On Wednesday 25 February 2015 11:04:16 Jonathan Richardson wrote:
> +- msr-override : array of strings to be used to override the individual
> + modem status signals for DCD, DSR, CTS, and RI. If the property is not
> + present, the individual signals are obtained from the modem status register.
> + Strings accepted are "dcd", "dsr", "cts", and "ri". If "dcd", "dsr", or
> + "cts" are present, these signals will always be reported as active. If
> + "ri" is present, this signal will always be reported as inactive.
>
Did you copy this definition from another driver? If not, we should try
to come up with a more natural way to do it. How about defining four
boolean properties instead, like
dcd-override
dsr-override
cts-override
ri-override
Simplifies the parser and the binding.
Arnd
Hi Arnd,
On 15-02-25 11:21 AM, Arnd Bergmann wrote:
> On Wednesday 25 February 2015 11:04:16 Jonathan Richardson wrote:
>> +- msr-override : array of strings to be used to override the individual
>> + modem status signals for DCD, DSR, CTS, and RI. If the property is not
>> + present, the individual signals are obtained from the modem status register.
>> + Strings accepted are "dcd", "dsr", "cts", and "ri". If "dcd", "dsr", or
>> + "cts" are present, these signals will always be reported as active. If
>> + "ri" is present, this signal will always be reported as inactive.
>>
>
> Did you copy this definition from another driver? If not, we should try
> to come up with a more natural way to do it. How about defining four
> boolean properties instead, like
>
> dcd-override
> dsr-override
> cts-override
> ri-override
>
> Simplifies the parser and the binding.
I didn't see anything similar in another driver. I agree with the bools
instead of strings. Did you want to keep these in the Synopsis driver
for now or were you thinking of putting them somewhere that they could
be re-used?
Thanks.
>
> Arnd
>
On Wednesday 25 February 2015 12:00:15 Jonathan Richardson wrote:
> On 15-02-25 11:21 AM, Arnd Bergmann wrote:
> > On Wednesday 25 February 2015 11:04:16 Jonathan Richardson wrote:
> >> +- msr-override : array of strings to be used to override the individual
> >> + modem status signals for DCD, DSR, CTS, and RI. If the property is not
> >> + present, the individual signals are obtained from the modem status register.
> >> + Strings accepted are "dcd", "dsr", "cts", and "ri". If "dcd", "dsr", or
> >> + "cts" are present, these signals will always be reported as active. If
> >> + "ri" is present, this signal will always be reported as inactive.
> >>
> >
> > Did you copy this definition from another driver? If not, we should try
> > to come up with a more natural way to do it. How about defining four
> > boolean properties instead, like
> >
> > dcd-override
> > dsr-override
> > cts-override
> > ri-override
> >
> > Simplifies the parser and the binding.
>
> I didn't see anything similar in another driver. I agree with the bools
> instead of strings. Did you want to keep these in the Synopsis driver
> for now or were you thinking of putting them somewhere that they could
> be re-used?
I'm fine with it either way. Unless someone else has an opinion on the
matter, feel free to pick what makes sense to you.
Arnd