[PATCH 4/4] hda: thinkpad_helper: Add support for hid-lenovo LED control

Subsystems: sound, the rest

STALE3691d

4 messages, 2 authors, 2016-09-14 · open the first message on its own page

[PATCH 4/4] hda: thinkpad_helper: Add support for hid-lenovo LED control

From: Dennis Wassenberg <hidden>
Date: 2016-09-12 10:47:03

Make the thinkpad_helper able to support not only led control over
acpi with thinkpad_acpi driver but also led control over hid-lenovo.
The hid-lenovo driver adapted the led control api of thinkpad_acpi.

Signed-off-by: Dennis Wassenberg <redacted>
---
 sound/pci/hda/thinkpad_helper.c | 149 ++++++++++++++++++++++++++++++----------
 1 file changed, 111 insertions(+), 38 deletions(-)
diff --git a/sound/pci/hda/thinkpad_helper.c b/sound/pci/hda/thinkpad_helper.c
index 62741a7..c24a4a9 100644
--- a/sound/pci/hda/thinkpad_helper.c
+++ b/sound/pci/hda/thinkpad_helper.c
@@ -2,79 +2,152 @@
  * to be included from codec driver
  */
 
-#if IS_ENABLED(CONFIG_THINKPAD_ACPI)
-
+#if IS_ENABLED(CONFIG_THINKPAD_ACPI) || IS_ENABLED(CONFIG_HID_LENOVO)
 #include <linux/acpi.h>
+#include <linux/hid-lenovo.h>
 #include <linux/thinkpad_acpi.h>
 
-static int (*led_set_func)(int, bool);
+static int (*led_set_func_tpacpi)(int, bool);
+static int (*led_set_func_hid_lenovo)(int, bool);
 static void (*old_vmaster_hook)(void *, int);
 
 static bool is_thinkpad(struct hda_codec *codec)
 {
+	return (codec->core.subsystem_id >> 16 == 0x17aa);
+}
+
+static bool is_thinkpad_acpi(struct hda_codec *codec)
+{
 	return (codec->core.subsystem_id >> 16 == 0x17aa) &&
 	       (acpi_dev_found("LEN0068") || acpi_dev_found("IBM0068"));
 }
 
-static void update_tpacpi_mute_led(void *private_data, int enabled)
+static void update_thinkpad_mute_led(void *private_data, int enabled)
 {
 	if (old_vmaster_hook)
 		old_vmaster_hook(private_data, enabled);
 
-	if (led_set_func)
-		led_set_func(TPACPI_LED_MUTE, !enabled);
+	if (led_set_func_tpacpi)
+		led_set_func_tpacpi(TPACPI_LED_MUTE, !enabled);
+
+	if (led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, !enabled);
 }
 
-static void update_tpacpi_micmute_led(struct hda_codec *codec,
+
+
+static void update_thinkpad_micmute_led(struct hda_codec *codec,
 				      struct snd_kcontrol *kcontrol,
 				      struct snd_ctl_elem_value *ucontrol)
 {
-	if (!ucontrol || !led_set_func)
+	if (!ucontrol)
 		return;
 	if (strcmp("Capture Switch", ucontrol->id.name) == 0 && ucontrol->id.index == 0) {
 		/* TODO: How do I verify if it's a mono or stereo here? */
 		bool val = ucontrol->value.integer.value[0] || ucontrol->value.integer.value[1];
-		led_set_func(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_tpacpi)
+			led_set_func_tpacpi(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_hid_lenovo)
+			led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, !val);
 	}
 }
 
-static void hda_fixup_thinkpad(struct hda_codec *codec,
-				    const struct hda_fixup *fix, int action)
+static int hda_fixup_thinkpad_acpi(struct hda_codec *codec)
 {
 	struct hda_gen_spec *spec = codec->spec;
-	bool removefunc = false;
+	int ret = -ENXIO;
 
-	if (action == HDA_FIXUP_ACT_PROBE) {
-		if (!is_thinkpad(codec))
-			return;
-		if (!led_set_func)
-			led_set_func = symbol_request(tpacpi_led_set);
-		if (!led_set_func) {
-			codec_warn(codec,
-				   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
-			return;
-		}
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!is_thinkpad_acpi(codec))
+		return -ENODEV;
+	if (!led_set_func_tpacpi)
+		led_set_func_tpacpi = symbol_request(tpacpi_led_set);
+	if (!led_set_func_tpacpi) {
+		codec_warn(codec,
+			   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
+		return -ENOENT;
+	}
 
-		removefunc = true;
-		if (led_set_func(TPACPI_LED_MUTE, false) >= 0) {
-			old_vmaster_hook = spec->vmaster_mute.hook;
-			spec->vmaster_mute.hook = update_tpacpi_mute_led;
-			removefunc = false;
-		}
-		if (led_set_func(TPACPI_LED_MICMUTE, false) >= 0) {
-			if (spec->num_adc_nids > 1)
-				codec_dbg(codec,
-					  "Skipping micmute LED control due to several ADCs");
-			else {
-				spec->cap_sync_hook = update_tpacpi_micmute_led;
-				removefunc = false;
-			}
+	if (led_set_func_tpacpi(TPACPI_LED_MUTE, false) >= 0) {
+		old_vmaster_hook = spec->vmaster_mute.hook;
+		spec->vmaster_mute.hook = update_thinkpad_mute_led;
+		ret = 0;
+	}
+
+	if (led_set_func_tpacpi(TPACPI_LED_MICMUTE, false) >= 0) {
+		if (spec->num_adc_nids > 1)
+			codec_dbg(codec,
+				  "Skipping micmute LED control due to several ADCs");
+		else {
+			spec->cap_sync_hook = update_thinkpad_micmute_led;
+			ret = 0;
 		}
 	}
 
-	if (led_set_func && (action == HDA_FIXUP_ACT_FREE || removefunc)) {
+	return ret;
+}
+
+static int hda_fixup_thinkpad_hid(struct hda_codec *codec)
+{
+	struct hda_gen_spec *spec = codec->spec;
+	int ret = 0;
+
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo = symbol_request(hid_lenovo_led_set);
+	if (!led_set_func_hid_lenovo) {
+		codec_warn(codec,
+			   "Failed to find hid-lenovo symbol hid_lenovo_led_set\n");
+		return -ENOENT;
+	}
+
+	if (update_thinkpad_mute_led != spec->vmaster_mute.hook)
+		old_vmaster_hook = spec->vmaster_mute.hook;
+
+	// do not remove hook if setting delay does not work currently because
+	// it is a usb hid devices which is not connected right now
+	// maybe is will be connected later
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, false);
+	spec->vmaster_mute.hook = update_thinkpad_mute_led;
+
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, false);
+	spec->cap_sync_hook = update_thinkpad_micmute_led;
+
+	return ret;
+}
+
+static void hda_fixup_thinkpad(struct hda_codec *codec,
+				    const struct hda_fixup *fix, int action)
+{
+	int ret_fixup_acpi = 0;
+	int ret_fixup_hid = 0;
+	bool remove = 0;
+
+	if (action == HDA_FIXUP_ACT_PROBE) {
+		ret_fixup_acpi = hda_fixup_thinkpad_acpi(codec);
+		ret_fixup_hid = hda_fixup_thinkpad_hid(codec);
+	}
+
+	if (led_set_func_tpacpi &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_acpi)) {
+
 		symbol_put(tpacpi_led_set);
-		led_set_func = NULL;
+		remove = true;
+	}
+
+	if (led_set_func_hid_lenovo &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_hid)) {
+
+		symbol_put(hid_lenovo_led_set);
+		remove = true;
+	}
+
+
+	if (remove) {
+		led_set_func_tpacpi = NULL;
+		led_set_func_hid_lenovo = NULL;
 		old_vmaster_hook = NULL;
 	}
 }
-- 
1.9.1

Re: [PATCH 4/4] hda: thinkpad_helper: Add support for hid-lenovo LED control

From: Takashi Iwai <hidden>
Date: 2016-09-12 12:38:14

On Mon, 12 Sep 2016 12:47:03 +0200,
Dennis Wassenberg wrote:
quoted hunk
Make the thinkpad_helper able to support not only led control over
acpi with thinkpad_acpi driver but also led control over hid-lenovo.
The hid-lenovo driver adapted the led control api of thinkpad_acpi.

Signed-off-by: Dennis Wassenberg <redacted>
---
 sound/pci/hda/thinkpad_helper.c | 149 ++++++++++++++++++++++++++++++----------
 1 file changed, 111 insertions(+), 38 deletions(-)
diff --git a/sound/pci/hda/thinkpad_helper.c b/sound/pci/hda/thinkpad_helper.c
index 62741a7..c24a4a9 100644
--- a/sound/pci/hda/thinkpad_helper.c
+++ b/sound/pci/hda/thinkpad_helper.c
@@ -2,79 +2,152 @@
  * to be included from codec driver
  */
 
-#if IS_ENABLED(CONFIG_THINKPAD_ACPI)
-
+#if IS_ENABLED(CONFIG_THINKPAD_ACPI) || IS_ENABLED(CONFIG_HID_LENOVO)
 #include <linux/acpi.h>
+#include <linux/hid-lenovo.h>
 #include <linux/thinkpad_acpi.h>
 
-static int (*led_set_func)(int, bool);
+static int (*led_set_func_tpacpi)(int, bool);
+static int (*led_set_func_hid_lenovo)(int, bool);
 static void (*old_vmaster_hook)(void *, int);
 
 static bool is_thinkpad(struct hda_codec *codec)
 {
+	return (codec->core.subsystem_id >> 16 == 0x17aa);
+}
+
+static bool is_thinkpad_acpi(struct hda_codec *codec)
+{
 	return (codec->core.subsystem_id >> 16 == 0x17aa) &&
 	       (acpi_dev_found("LEN0068") || acpi_dev_found("IBM0068"));
 }
 
-static void update_tpacpi_mute_led(void *private_data, int enabled)
+static void update_thinkpad_mute_led(void *private_data, int enabled)
 {
 	if (old_vmaster_hook)
 		old_vmaster_hook(private_data, enabled);
 
-	if (led_set_func)
-		led_set_func(TPACPI_LED_MUTE, !enabled);
+	if (led_set_func_tpacpi)
+		led_set_func_tpacpi(TPACPI_LED_MUTE, !enabled);
+
+	if (led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, !enabled);
 }
 
-static void update_tpacpi_micmute_led(struct hda_codec *codec,
+
+
+static void update_thinkpad_micmute_led(struct hda_codec *codec,
 				      struct snd_kcontrol *kcontrol,
 				      struct snd_ctl_elem_value *ucontrol)
 {
-	if (!ucontrol || !led_set_func)
+	if (!ucontrol)
 		return;
 	if (strcmp("Capture Switch", ucontrol->id.name) == 0 && ucontrol->id.index == 0) {
 		/* TODO: How do I verify if it's a mono or stereo here? */
 		bool val = ucontrol->value.integer.value[0] || ucontrol->value.integer.value[1];
-		led_set_func(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_tpacpi)
+			led_set_func_tpacpi(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_hid_lenovo)
+			led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, !val);
 	}
 }
 
-static void hda_fixup_thinkpad(struct hda_codec *codec,
-				    const struct hda_fixup *fix, int action)
+static int hda_fixup_thinkpad_acpi(struct hda_codec *codec)
 {
 	struct hda_gen_spec *spec = codec->spec;
-	bool removefunc = false;
+	int ret = -ENXIO;
 
-	if (action == HDA_FIXUP_ACT_PROBE) {
-		if (!is_thinkpad(codec))
-			return;
-		if (!led_set_func)
-			led_set_func = symbol_request(tpacpi_led_set);
-		if (!led_set_func) {
-			codec_warn(codec,
-				   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
-			return;
-		}
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!is_thinkpad_acpi(codec))
+		return -ENODEV;
+	if (!led_set_func_tpacpi)
+		led_set_func_tpacpi = symbol_request(tpacpi_led_set);
This would be performed even if CONFIG_THINKPAD_ACPI=n when
CONFIG_HID_LENOVO!=n.  You'd need to have a proper ifdef surrounding
the function.

quoted hunk
+	if (!led_set_func_tpacpi) {
+		codec_warn(codec,
+			   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
+		return -ENOENT;
+	}
 
-		removefunc = true;
-		if (led_set_func(TPACPI_LED_MUTE, false) >= 0) {
-			old_vmaster_hook = spec->vmaster_mute.hook;
-			spec->vmaster_mute.hook = update_tpacpi_mute_led;
-			removefunc = false;
-		}
-		if (led_set_func(TPACPI_LED_MICMUTE, false) >= 0) {
-			if (spec->num_adc_nids > 1)
-				codec_dbg(codec,
-					  "Skipping micmute LED control due to several ADCs");
-			else {
-				spec->cap_sync_hook = update_tpacpi_micmute_led;
-				removefunc = false;
-			}
+	if (led_set_func_tpacpi(TPACPI_LED_MUTE, false) >= 0) {
+		old_vmaster_hook = spec->vmaster_mute.hook;
+		spec->vmaster_mute.hook = update_thinkpad_mute_led;
+		ret = 0;
+	}
+
+	if (led_set_func_tpacpi(TPACPI_LED_MICMUTE, false) >= 0) {
+		if (spec->num_adc_nids > 1)
+			codec_dbg(codec,
+				  "Skipping micmute LED control due to several ADCs");
+		else {
+			spec->cap_sync_hook = update_thinkpad_micmute_led;
+			ret = 0;
 		}
 	}
 
-	if (led_set_func && (action == HDA_FIXUP_ACT_FREE || removefunc)) {
+	return ret;
+}
+
+static int hda_fixup_thinkpad_hid(struct hda_codec *codec)
+{
+	struct hda_gen_spec *spec = codec->spec;
+	int ret = 0;
+
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo = symbol_request(hid_lenovo_led_set);
+	if (!led_set_func_hid_lenovo) {
+		codec_warn(codec,
+			   "Failed to find hid-lenovo symbol hid_lenovo_led_set\n");
+		return -ENOENT;
+	}
+
+	if (update_thinkpad_mute_led != spec->vmaster_mute.hook)
+		old_vmaster_hook = spec->vmaster_mute.hook;
+
+	// do not remove hook if setting delay does not work currently because
+	// it is a usb hid devices which is not connected right now
+	// maybe is will be connected later
The comment should be C style.  Try checkpatch.pl once before the
patch submission.

quoted hunk
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, false);
+	spec->vmaster_mute.hook = update_thinkpad_mute_led;
+
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, false);
+	spec->cap_sync_hook = update_thinkpad_micmute_led;
+
+	return ret;
+}
+
+static void hda_fixup_thinkpad(struct hda_codec *codec,
+				    const struct hda_fixup *fix, int action)
+{
+	int ret_fixup_acpi = 0;
+	int ret_fixup_hid = 0;
+	bool remove = 0;
+
+	if (action == HDA_FIXUP_ACT_PROBE) {
+		ret_fixup_acpi = hda_fixup_thinkpad_acpi(codec);
+		ret_fixup_hid = hda_fixup_thinkpad_hid(codec);
Basically these two are exclusive.  So you don't have to process the
latter when the former succeeds.  But...
quoted hunk
+	}
+
+	if (led_set_func_tpacpi &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_acpi)) {
+
 		symbol_put(tpacpi_led_set);
-		led_set_func = NULL;
+		remove = true;
+	}
+
+	if (led_set_func_hid_lenovo &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_hid)) {
+
+		symbol_put(hid_lenovo_led_set);
+		remove = true;
+	}
+
+
+	if (remove) {
+		led_set_func_tpacpi = NULL;
+		led_set_func_hid_lenovo = NULL;
 		old_vmaster_hook = NULL;
 	}
 }
... reading through the patch, I find the code is a bit too
redundant.  The access is exclusive between ACPI and HID, and the hook
functions are in the same form but just the argument value is
different (TPACPI_LED_MUTE vs HID_LENOVO_LED_MUTE).  So we may need
only a single function pointer and pass the different value depending
on the model instead of keeping two individual function pointers.


thanks,

Takashi

Re: [PATCH 4/4] hda: thinkpad_helper: Add support for hid-lenovo LED control

From: Dennis Wassenberg <hidden>
Date: 2016-09-14 06:23:54

Hi Takashi,

at my first shot published to github I had an approach which makes thinkpad_acpi and hid-lenovo mutual exclusive (https://github.com/dwassenberg/linux/tree/development/hid-rmi). This approach has some issues. For example in case of thinkpad_acpi and hid-lenovo module is loaded (e.g. at Lenovo ThinkPad X1 Tablet) the symbol request to tpacpi_led_set is successful so the thinkpad_helper will decide in favor of thinkpad_acpi. So I need to add something which checks the return code of tpacpi_led_set. If -ENODEV is returned it can try hid-lenovo.

But the reason why I made thinkpad_acpi and hid-lenovo work in parallel was that if it is mutual exclusive there is no possibility to control LEDs of external Lenovo USB keyboards which are connected to a Thinkpad. Currently (including the other patches) there is only the X1 Tablet Cover LEDs which is controllable from external code. But the hid-lenovo driver is able to handle different Lenovo USB Keyboards.

- ThinkPad USB Keyboard with TrackPoint (tpkbd)
- ThinkPad Compact Bluetooth Keyboard with TrackPoint (cptkbd)
- ThinkPad Compact USB Keyboard with TrackPoint (cptkbd)
- ThinkPad X1 Cover USB Keyboard with TrackPoint and Touchpad (tpx1cover)

These keyboards have these LEDs too. For some of these keyboards a LED control is implemented inside hid-lenovo too. If thinkpad_acpi and hid-lenovo is mutual exclusive it is impossible to let the thinkpad_helper control the LEDs of these external USB keyboards if they are connected to a thinkpad. I don't know if anyone will do this, but I don't want to take the possibility to do such thinks :)

So.. if there is the decision (from your side) that this should be mutual exclusive I will prepare a v2 which the mutual exclusive approach I described at the top.

Thank you & best regards,

Dennis

On 12.09.2016 14:38, Takashi Iwai wrote:
On Mon, 12 Sep 2016 12:47:03 +0200,
Dennis Wassenberg wrote:
quoted
Make the thinkpad_helper able to support not only led control over
acpi with thinkpad_acpi driver but also led control over hid-lenovo.
The hid-lenovo driver adapted the led control api of thinkpad_acpi.

Signed-off-by: Dennis Wassenberg <redacted>
---
 sound/pci/hda/thinkpad_helper.c | 149 ++++++++++++++++++++++++++++++----------
 1 file changed, 111 insertions(+), 38 deletions(-)
diff --git a/sound/pci/hda/thinkpad_helper.c b/sound/pci/hda/thinkpad_helper.c
index 62741a7..c24a4a9 100644
--- a/sound/pci/hda/thinkpad_helper.c
+++ b/sound/pci/hda/thinkpad_helper.c
@@ -2,79 +2,152 @@
  * to be included from codec driver
  */
 
-#if IS_ENABLED(CONFIG_THINKPAD_ACPI)
-
+#if IS_ENABLED(CONFIG_THINKPAD_ACPI) || IS_ENABLED(CONFIG_HID_LENOVO)
 #include <linux/acpi.h>
+#include <linux/hid-lenovo.h>
 #include <linux/thinkpad_acpi.h>
 
-static int (*led_set_func)(int, bool);
+static int (*led_set_func_tpacpi)(int, bool);
+static int (*led_set_func_hid_lenovo)(int, bool);
 static void (*old_vmaster_hook)(void *, int);
 
 static bool is_thinkpad(struct hda_codec *codec)
 {
+	return (codec->core.subsystem_id >> 16 == 0x17aa);
+}
+
+static bool is_thinkpad_acpi(struct hda_codec *codec)
+{
 	return (codec->core.subsystem_id >> 16 == 0x17aa) &&
 	       (acpi_dev_found("LEN0068") || acpi_dev_found("IBM0068"));
 }
 
-static void update_tpacpi_mute_led(void *private_data, int enabled)
+static void update_thinkpad_mute_led(void *private_data, int enabled)
 {
 	if (old_vmaster_hook)
 		old_vmaster_hook(private_data, enabled);
 
-	if (led_set_func)
-		led_set_func(TPACPI_LED_MUTE, !enabled);
+	if (led_set_func_tpacpi)
+		led_set_func_tpacpi(TPACPI_LED_MUTE, !enabled);
+
+	if (led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, !enabled);
 }
 
-static void update_tpacpi_micmute_led(struct hda_codec *codec,
+
+
+static void update_thinkpad_micmute_led(struct hda_codec *codec,
 				      struct snd_kcontrol *kcontrol,
 				      struct snd_ctl_elem_value *ucontrol)
 {
-	if (!ucontrol || !led_set_func)
+	if (!ucontrol)
 		return;
 	if (strcmp("Capture Switch", ucontrol->id.name) == 0 && ucontrol->id.index == 0) {
 		/* TODO: How do I verify if it's a mono or stereo here? */
 		bool val = ucontrol->value.integer.value[0] || ucontrol->value.integer.value[1];
-		led_set_func(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_tpacpi)
+			led_set_func_tpacpi(TPACPI_LED_MICMUTE, !val);
+		if (led_set_func_hid_lenovo)
+			led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, !val);
 	}
 }
 
-static void hda_fixup_thinkpad(struct hda_codec *codec,
-				    const struct hda_fixup *fix, int action)
+static int hda_fixup_thinkpad_acpi(struct hda_codec *codec)
 {
 	struct hda_gen_spec *spec = codec->spec;
-	bool removefunc = false;
+	int ret = -ENXIO;
 
-	if (action == HDA_FIXUP_ACT_PROBE) {
-		if (!is_thinkpad(codec))
-			return;
-		if (!led_set_func)
-			led_set_func = symbol_request(tpacpi_led_set);
-		if (!led_set_func) {
-			codec_warn(codec,
-				   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
-			return;
-		}
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!is_thinkpad_acpi(codec))
+		return -ENODEV;
+	if (!led_set_func_tpacpi)
+		led_set_func_tpacpi = symbol_request(tpacpi_led_set);
This would be performed even if CONFIG_THINKPAD_ACPI=n when
CONFIG_HID_LENOVO!=n.  You'd need to have a proper ifdef surrounding
the function.

quoted
+	if (!led_set_func_tpacpi) {
+		codec_warn(codec,
+			   "Failed to find thinkpad-acpi symbol tpacpi_led_set\n");
+		return -ENOENT;
+	}
 
-		removefunc = true;
-		if (led_set_func(TPACPI_LED_MUTE, false) >= 0) {
-			old_vmaster_hook = spec->vmaster_mute.hook;
-			spec->vmaster_mute.hook = update_tpacpi_mute_led;
-			removefunc = false;
-		}
-		if (led_set_func(TPACPI_LED_MICMUTE, false) >= 0) {
-			if (spec->num_adc_nids > 1)
-				codec_dbg(codec,
-					  "Skipping micmute LED control due to several ADCs");
-			else {
-				spec->cap_sync_hook = update_tpacpi_micmute_led;
-				removefunc = false;
-			}
+	if (led_set_func_tpacpi(TPACPI_LED_MUTE, false) >= 0) {
+		old_vmaster_hook = spec->vmaster_mute.hook;
+		spec->vmaster_mute.hook = update_thinkpad_mute_led;
+		ret = 0;
+	}
+
+	if (led_set_func_tpacpi(TPACPI_LED_MICMUTE, false) >= 0) {
+		if (spec->num_adc_nids > 1)
+			codec_dbg(codec,
+				  "Skipping micmute LED control due to several ADCs");
+		else {
+			spec->cap_sync_hook = update_thinkpad_micmute_led;
+			ret = 0;
 		}
 	}
 
-	if (led_set_func && (action == HDA_FIXUP_ACT_FREE || removefunc)) {
+	return ret;
+}
+
+static int hda_fixup_thinkpad_hid(struct hda_codec *codec)
+{
+	struct hda_gen_spec *spec = codec->spec;
+	int ret = 0;
+
+	if (!is_thinkpad(codec))
+		return -ENODEV;
+	if (!led_set_func_hid_lenovo)
+		led_set_func_hid_lenovo = symbol_request(hid_lenovo_led_set);
+	if (!led_set_func_hid_lenovo) {
+		codec_warn(codec,
+			   "Failed to find hid-lenovo symbol hid_lenovo_led_set\n");
+		return -ENOENT;
+	}
+
+	if (update_thinkpad_mute_led != spec->vmaster_mute.hook)
+		old_vmaster_hook = spec->vmaster_mute.hook;
+
+	// do not remove hook if setting delay does not work currently because
+	// it is a usb hid devices which is not connected right now
+	// maybe is will be connected later
The comment should be C style.  Try checkpatch.pl once before the
patch submission.

quoted
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MUTE, false);
+	spec->vmaster_mute.hook = update_thinkpad_mute_led;
+
+	led_set_func_hid_lenovo(HID_LENOVO_LED_MICMUTE, false);
+	spec->cap_sync_hook = update_thinkpad_micmute_led;
+
+	return ret;
+}
+
+static void hda_fixup_thinkpad(struct hda_codec *codec,
+				    const struct hda_fixup *fix, int action)
+{
+	int ret_fixup_acpi = 0;
+	int ret_fixup_hid = 0;
+	bool remove = 0;
+
+	if (action == HDA_FIXUP_ACT_PROBE) {
+		ret_fixup_acpi = hda_fixup_thinkpad_acpi(codec);
+		ret_fixup_hid = hda_fixup_thinkpad_hid(codec);
Basically these two are exclusive.  So you don't have to process the
latter when the former succeeds.  But...
quoted
+	}
+
+	if (led_set_func_tpacpi &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_acpi)) {
+
 		symbol_put(tpacpi_led_set);
-		led_set_func = NULL;
+		remove = true;
+	}
+
+	if (led_set_func_hid_lenovo &&
+		(action == HDA_FIXUP_ACT_FREE || ret_fixup_hid)) {
+
+		symbol_put(hid_lenovo_led_set);
+		remove = true;
+	}
+
+
+	if (remove) {
+		led_set_func_tpacpi = NULL;
+		led_set_func_hid_lenovo = NULL;
 		old_vmaster_hook = NULL;
 	}
 }
... reading through the patch, I find the code is a bit too
redundant.  The access is exclusive between ACPI and HID, and the hook
functions are in the same form but just the argument value is
different (TPACPI_LED_MUTE vs HID_LENOVO_LED_MUTE).  So we may need
only a single function pointer and pass the different value depending
on the model instead of keeping two individual function pointers.


thanks,

Takashi

Re: [PATCH 4/4] hda: thinkpad_helper: Add support for hid-lenovo LED control

From: Takashi Iwai <hidden>
Date: 2016-09-14 07:03:50

On Wed, 14 Sep 2016 08:22:45 +0200,
Dennis Wassenberg wrote:
Hi Takashi,

at my first shot published to github I had an approach which makes thinkpad_acpi and hid-lenovo mutual exclusive (https://github.com/dwassenberg/linux/tree/development/hid-rmi). This approach has some issues. For example in case of thinkpad_acpi and hid-lenovo module is loaded (e.g. at Lenovo ThinkPad X1 Tablet) the symbol request to tpacpi_led_set is successful so the thinkpad_helper will decide in favor of thinkpad_acpi. So I need to add something which checks the return code of tpacpi_led_set. If -ENODEV is returned it can try hid-lenovo.

But the reason why I made thinkpad_acpi and hid-lenovo work in parallel was that if it is mutual exclusive there is no possibility to control LEDs of external Lenovo USB keyboards which are connected to a Thinkpad. Currently (including the other patches) there is only the X1 Tablet Cover LEDs which is controllable from external code. But the hid-lenovo driver is able to handle different Lenovo USB Keyboards.

- ThinkPad USB Keyboard with TrackPoint (tpkbd)
- ThinkPad Compact Bluetooth Keyboard with TrackPoint (cptkbd)
- ThinkPad Compact USB Keyboard with TrackPoint (cptkbd)
- ThinkPad X1 Cover USB Keyboard with TrackPoint and Touchpad (tpx1cover)

These keyboards have these LEDs too. For some of these keyboards a LED control is implemented inside hid-lenovo too. If thinkpad_acpi and hid-lenovo is mutual exclusive it is impossible to let the thinkpad_helper control the LEDs of these external USB keyboards if they are connected to a thinkpad. I don't know if anyone will do this, but I don't want to take the possibility to do such thinks :)

So.. if there is the decision (from your side) that this should be mutual exclusive I will prepare a v2 which the mutual exclusive approach I described at the top.
OK, now understood.  But still make sure that it works even with one
of kconfigs is disabled.


thanks,

Takashi
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help