Skip to contentSkip to conversation

leds: ledtrig-tty: add tty_led_mode xtension

PATCH In review Florian Eckert 2 participants 6 messages Mar 6, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v8 cover leds: ledtrig-tty: add tty_led_mode xtension

Hello,

here commes v8 of this series to add additional tty_led_modes.

v8:
Changes compared to the v7 patchset with
[email protected] are.

* Update 'Documentation/ABI/testing/sysfs-class-led-trigger-tty' with the suggested changes.
* Use a meaningfull name for the Rx/Tx enum.
* Do not use C++ comments style

Add a new patch 'ledtrig-tty-fix-brightness-set' to fix an issue with LEDs that are connected via GPIOs to the SOC with the device driver 'leds-gpio'. In my view, the wrong function is being used here. LEDs that do not need and do not have a delayed call do not work with the current implementation. Therefore, the 'led_set_brightness_nosleep' function must be used.

v7:
Changes compared to the v5 patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

Thanks for the hint with the command 'make htmldocs SPHINXDIRS="admin-guide"'. Unfortunately, I did not know that. I have now verified it also in the browser. In my opinion, the list is now also displayed correctly in the documentation.

v6:
Changes compared to the v5 patchset with
[email protected] are.

Addressed review comments by kernel test robot are:

* fix Documentation/ABI/testing/sysfs-class-led-trigger-tty:9:
WARNING: Unexpected indentation.
* fix Documentation/ABI/testing/sysfs-class-led-trigger-tty:9:
WARNING: Block quote ends without a blank line; unexpected unindent.

Thanks to Jiri Slaby, who gave me the crucial hint of what I need to fix, to possibly make the 'Kernel test robot' happy.

v5:
Changes compared to the v4 patchset with
[email protected] are.

Sorry for the inconvenience, but I sent the wrong patch for ledtrig-tty.c in v4. The v5 patchset now includes all the changes I specified in the v4 patchset.

v4:
Changes compared to the v3 patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

ledtrig-tty.c:
- Do not use __TTY_LED_MAX pattern us instead __TTY_LED_LAST = TTY_LED_RNG
- Move declartion and assignment into one singel line
- Use __TTY_LED_LAST pattern, to simplify tty_mode_show and tty_mode_store handling

v3:
Changes compared to the v2 patchset with
[email protected] are.

Addressed review comments by Greg K-H are:

tty.h:
- Fix first comment line and remark -%ENOTTY for the new function 'tty_get_mget' to make a proper kernel doc.
- Add the return value -%ENOTTY again, I thought it was no longer needed.

v2:
Changes compared to the initial patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

tty.h:
- Fix compilation error because of wrong rebaseing
- Remove empty lines
- Use new 'tty_get_mget' in 'tty_tiocmget'

ledtrig-tty.c:
- Update commit description
- Use enum for tty_led_mod in struct ledtrig_tty_date
- Rename sysfs file from 'mode' to 'tty_led_mode'
- Change tty_led_mode show function to use loop instead of switch/case
- Change tty_led_mode store function to use loop instead of switch/case
- Check return value of function tty_get_mget

Florian Eckert (3):
tty: new helper function tty_get_mget
trigger: ledtrig-tty: add additional modes
ledtrig-tty: call correct brightness set function

 .../ABI/testing/sysfs-class-led-trigger-tty   |  18 +++
 drivers/leds/trigger/ledtrig-tty.c            | 145 ++++++++++++++++--
 drivers/tty/tty_io.c                          |  28 +++-
 include/linux/tty.h                           |   1 +
 4 files changed, 171 insertions(+), 21 deletions(-)
Signature · 1 line
2.30.2
#1 lore raw

PATCH v8 1/3 tty: new helper function tty_get_mget

For a given struct tty_struct, this provides the appropriate tty line state flags needed to add more modes to the ledtrig-tty trigger.

The new function is then used to get via tty_tiocmget() the different tty line states.

---
 drivers/tty/tty_io.c | 28 ++++++++++++++++++++++------
 include/linux/tty.h  |  1 +
 2 files changed, 23 insertions(+), 6 deletions(-)
drivers/tty/tty_io.c +22 -6
index 3149114bf130..a068b03a0828 100644
@@ -2493,6 +2493,24 @@ static int send_break(struct tty_struct *tty, unsigned int duration)
 	return retval;
 }
 
+/**
+ * tty_get_mget		-	get modem status
+ * @tty: tty device
+ *
+ * Obtain the modem status bits from the tty driver if the feature
+ * is supported. Return -%ENOTTY if it is not available.
+ */
+int tty_get_mget(struct tty_struct *tty)
+{
+	int retval = -ENOTTY;
+
+	if (tty->ops->tiocmget)
+		retval = tty->ops->tiocmget(tty);
+
+	return retval;
+}
+EXPORT_SYMBOL_GPL(tty_get_mget);
+
 /**
  * tty_tiocmget		-	get modem status
  * @tty: tty device
@@ -2505,14 +2523,12 @@ static int send_break(struct tty_struct *tty, unsigned int duration)
  */
 static int tty_tiocmget(struct tty_struct *tty, int __user *p)
 {
-	int retval = -ENOTTY;
+	int retval;
 
-	if (tty->ops->tiocmget) {
-		retval = tty->ops->tiocmget(tty);
+	retval = tty_get_mget(tty);
+	if (retval >= 0)
+		retval = put_user(retval, p);
 
-		if (retval >= 0)
-			retval = put_user(retval, p);
-	}
 	return retval;
 }
 
include/linux/tty.h +1 -0
index 730c3301d710..825186c0fec1 100644
@@ -421,6 +421,7 @@ int tty_unthrottle_safe(struct tty_struct *tty);
 int tty_do_resize(struct tty_struct *tty, struct winsize *ws);
 int tty_get_icount(struct tty_struct *tty,
 		struct serial_icounter_struct *icount);
+int tty_get_mget(struct tty_struct *tty);
 int is_current_pgrp_orphaned(void);
 void tty_hangup(struct tty_struct *tty);
 void tty_vhangup(struct tty_struct *tty);
Signature · 1 line
2.30.2
#2 lore raw

PATCH v8 2/3 trigger: ledtrig-tty: add additional modes

In reply to Florian Eckert

Add additional modes to trigger the selected LED.
The following modes are supported:

Tx/Rx: Flash LED on data transmission (default)
CTS: DCE Ready to accept data from the DTE.
DSR: DCE is ready to receive and send data.
CAR: DCE is receiving a carrier from a remote DTE.
RNG: DCE has detected an incoming ring signal.

The mode can be changed for example with the following command:
echo "CTS" > /sys/class/leds/<led>/mode

This would turn on the LED, when the DTE(modem) signals the DCE that it is ready to accept data.

---
 .../ABI/testing/sysfs-class-led-trigger-tty   |  18 +++
 drivers/leds/trigger/ledtrig-tty.c            | 143 ++++++++++++++++--
 2 files changed, 146 insertions(+), 15 deletions(-)
Documentation/ABI/testing/sysfs-class-led-trigger-tty +18 -0
index 2bf6b24e781b..2e4e4c009a21 100644
@@ -4,3 +4,21 @@ KernelVersion:	5.10
 Contact:	[email protected]
 Description:
 		Specifies the tty device name of the triggering tty
+
+What:		/sys/class/leds/<led>/mode
+Date:		January 2023
+KernelVersion:	6.3
+Description:
+		Specifies the mode to trigger the LED.
+		The following operating modes are supported:
+
+		=====  ============================================
+		Tx/Rx  Flash LED on data transmission (default)
+		CTS    DCE Ready to accept data from the DTE
+		DSR    DCE is ready to receive and send data
+		CAR    DCE is receiving a carrier from a remote DTE
+		RNG    DCE has detected an incoming ring signal
+		=====  ============================================
+
+		In CTS/DSR/CAR/RNG modes, the LED is enabled when the
+		tty line is high.
\ No newline at end of file
drivers/leds/trigger/ledtrig-tty.c +128 -15

185 lines not on this page · raw message

Signature · 1 line
2.30.2
#3 lore raw

PATCH v8 3/3 ledtrig-tty: call correct brightness set function

In reply to Florian Eckert

The problem is, however, that with LEDs that are connected directly to the SOC via a GPIO, the LED cannot be switched on or off. Because the 'leds-gpio' device driver does not need a delayed work.

If we look at the callstack for the function 'led_set_brightness_sync', then at the end it is checked whether the LED device driver has the function 'brightness_set_blocking' implemented. This is not the case for the 'leds-gpio' device.

The function 'brightness_set_blocking' returns LED devices -ENOTSUPP. This is not evaluated in 'ledtrig-tty' trigger.

In order for the trigger to also work with LED that does not have the 'brightness_set_blocking' function implemented, we must therefore use the internal API of the LED subsystem and use the function 'led_set_brightness_nosleep'.

---
 drivers/leds/trigger/ledtrig-tty.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/leds/trigger/ledtrig-tty.c +6 -4
index d3407830fa69..132e809898a7 100644
@@ -7,6 +7,8 @@
 #include <linux/tty.h>
 #include <uapi/linux/serial.h>
 
+#include "../leds.h"
+
 enum tty_led_mode {
 	TTY_LED_RXTX,
 	TTY_LED_CTS,
@@ -164,9 +166,9 @@ static void ledtrig_tty_flags(struct ledtrig_tty_data *trigger_data, unsigned in
 
 	status = tty_get_mget(trigger_data->tty);
 	if (status & flag)
-		led_set_brightness_sync(trigger_data->led_cdev, LED_ON);
+		led_set_brightness_nosleep(trigger_data->led_cdev, LED_ON);
 	else
-		led_set_brightness_sync(trigger_data->led_cdev, LED_OFF);
+		led_set_brightness_nosleep(trigger_data->led_cdev, LED_OFF);
 }
 
 static void ledtrig_tty_work(struct work_struct *work)
@@ -229,12 +231,12 @@ static void ledtrig_tty_work(struct work_struct *work)
 		}
 
 		if (icount.rx != trigger_data->rx || icount.tx != trigger_data->tx) {
-			led_set_brightness_sync(trigger_data->led_cdev, LED_ON);
+			led_set_brightness_nosleep(trigger_data->led_cdev, LED_ON);
 
 			trigger_data->rx = icount.rx;
 			trigger_data->tx = icount.tx;
 		} else {
-			led_set_brightness_sync(trigger_data->led_cdev, LED_OFF);
+			led_set_brightness_nosleep(trigger_data->led_cdev, LED_OFF);
 		}
 		break;
 	}
Signature · 1 line
2.30.2
maintainer
#4 lore raw
In reply to Florian Eckert
Quoted 3 lines · Florian Eckert

Hello,

here commes v8 of this series to add additional tty_led_modes.

I think u r 2 fast. Uwe had some more comments on v7.

Don't be afraid to wait some days after the last comments ;). As it is, it makes the review harder...

Quoted 108 lines

v8:
Changes compared to the v7 patchset with
[email protected] are.

* Update 'Documentation/ABI/testing/sysfs-class-led-trigger-tty' with the suggested changes.
* Use a meaningfull name for the Rx/Tx enum.
* Do not use C++ comments style

Add a new patch 'ledtrig-tty-fix-brightness-set' to fix an issue with LEDs that are connected via GPIOs to the SOC with the device driver 'leds-gpio'. In my view, the wrong function is being used here. LEDs that do not need and do not have a delayed call do not work with the current implementation. Therefore, the 'led_set_brightness_nosleep' function must be used.

v7:
Changes compared to the v5 patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

Thanks for the hint with the command 'make htmldocs SPHINXDIRS="admin-guide"'. Unfortunately, I did not know that. I have now verified it also in the browser. In my opinion, the list is now also displayed correctly in the documentation.

v6:
Changes compared to the v5 patchset with
[email protected] are.

Addressed review comments by kernel test robot are:

* fix Documentation/ABI/testing/sysfs-class-led-trigger-tty:9:
WARNING: Unexpected indentation.
* fix Documentation/ABI/testing/sysfs-class-led-trigger-tty:9:
WARNING: Block quote ends without a blank line; unexpected unindent.

Thanks to Jiri Slaby, who gave me the crucial hint of what I need to fix, to possibly make the 'Kernel test robot' happy.

v5:
Changes compared to the v4 patchset with
[email protected] are.

Sorry for the inconvenience, but I sent the wrong patch for ledtrig-tty.c in v4. The v5 patchset now includes all the changes I specified in the v4 patchset.

v4:
Changes compared to the v3 patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

ledtrig-tty.c:
- Do not use __TTY_LED_MAX pattern us instead __TTY_LED_LAST = TTY_LED_RNG
- Move declartion and assignment into one singel line
- Use __TTY_LED_LAST pattern, to simplify tty_mode_show and tty_mode_store handling

v3:
Changes compared to the v2 patchset with
[email protected] are.

Addressed review comments by Greg K-H are:

tty.h:
- Fix first comment line and remark -%ENOTTY for the new function 'tty_get_mget' to make a proper kernel doc.
- Add the return value -%ENOTTY again, I thought it was no longer needed.

v2:
Changes compared to the initial patchset with
[email protected] are.

Addressed review comments by Jiri Slaby are:

tty.h:
- Fix compilation error because of wrong rebaseing
- Remove empty lines
- Use new 'tty_get_mget' in 'tty_tiocmget'

ledtrig-tty.c:
- Update commit description
- Use enum for tty_led_mod in struct ledtrig_tty_date
- Rename sysfs file from 'mode' to 'tty_led_mode'
- Change tty_led_mode show function to use loop instead of switch/case
- Change tty_led_mode store function to use loop instead of switch/case
- Check return value of function tty_get_mget

Florian Eckert (3):
tty: new helper function tty_get_mget
trigger: ledtrig-tty: add additional modes
ledtrig-tty: call correct brightness set function

  .../ABI/testing/sysfs-class-led-trigger-tty   |  18 +++
  drivers/leds/trigger/ledtrig-tty.c            | 145 ++++++++++++++++--
  drivers/tty/tty_io.c                          |  28 +++-
  include/linux/tty.h                           |   1 +
  4 files changed, 171 insertions(+), 21 deletions(-)
Signature · 2 lines
js
suse labs
#5 lore raw
Quoted 9 lines
Quoted 3 lines

here commes v8 of this series to add additional tty_led_modes.

I think u r 2 fast. Uwe had some more comments on v7.

Don't be afraid to wait some days after the last comments ;). As it is, it makes the review harder...

That was probably a race condition ;-) This must have been received when I sent the v8.