leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false
PATCH 1/1 leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);
the props variable is still initialized to zero.
Call the BUG_ON only after we parse fwnode into props.
--- drivers/leds/led-core.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/leds/led-core.c +4 -4
@@ -474,15 +474,15 @@ int led_compose_name(struct device *dev, struct led_init_data *init_data,struct fwnode_handle *fwnode = init_data->fwnode;const char *devicename = init_data->devicename; -/* We want to label LEDs that can produce full range of colors -* as RGB, not multicolor */ -BUG_ON (props.color ==LED_COLOR_ID_MULTI ); -if (!led_classdev_name)return -EINVAL ;led_parse_fwnode_props (dev, fwnode, &props); +/* We want to label LEDs that can produce full range of colors +* as RGB, not multicolor */ +BUG_ON (props.color ==LED_COLOR_ID_MULTI ); +if (props.label) {/* * If init_data.devicename is NULL, then it indicates that
Signature · 1 line
2.41.0
Re: (subset) [PATCH] leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false
Quoted 8 lines · Marek Behún
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);the props variable is still initialized to zero.
Call the
BUG_ONonly after we parse fwnode into props.[...]
Applied, thanks!
[1/1] leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false
commit: c3f853184bed04105682383c2971798c572226b5
--
Lee Jones [李琼斯]
Quoted 8 lines · Marek Beh'un
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);the props variable is still initialized to zero.
Call the
BUG_ONonly after we parse fwnode into props.Fixes 77dce3a22e89 ("leds: disallow /sys/class/leds/*:multi:* for now")Signed-off-by Marek Beh'un <[email protected]>
I've just discovered this has broken boot on my Libre Computer AML-A311D-CC-V0.2, which was working just fine with Debian 12's stock kernel:
mark@stor-flodeboller:~$ uname -a Linux stor-flodeboller 6.1.0-12-arm64 #1 SMP Debian 6.1.52-1 (2023-09-07) aarch64 GNU/Linux
When upgrading to v6.6-rc3 for testing, the board dies at boot time due to the BUG_ON() moved by this commit. That BUG_ON() provides no useful context for solving the issue, and it's *distinctly* unhelpful.
I decompiled the DTB provided by the firmware, and it has:
| leds {
| compatible = "gpio-leds";
|
| led-yellow-green {
| color = <0x08>;
| function = "status";
| gpios = <0x16 0x44 0x01>;
| linux,default-trigger = "default-on";
| panic-indicator;
| };
|
| led-blue {
| color = <0x03>;
| function = "activity";
| gpios = <0x16 0x48 0x01>;
| linux,default-trigger = "activity";
| };
| };... and IIUC LED_COLOR_ID_MULTI is 8, so the problem is the led-yellow-green subnode. I presume that could use LED_COLOR_ID_RGB.
Can we please do something so that this doesn't prevent the board from booting?
Is there some reason we don't transparently convert LED_COLOR_ID_MULTI to LED_COLOR_ID_RGB?
If that's not viable, could we skip the LED *without* crashing the kernel, so that the board remains usable?
Thanks,
Mark.
Quoted 32 lines
--- drivers/leds/led-core.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)drivers/leds/led-core.c +4
-4@@ -474,15 +474,15 @@ int led_compose_name(struct device *dev, struct led_init_data *init_data,struct fwnode_handle *fwnode = init_data->fwnode;const char *devicename = init_data->devicename; -/* We want to label LEDs that can produce full range of colors -* as RGB, not multicolor */ -BUG_ON (props.color ==LED_COLOR_ID_MULTI ); -if (!led_classdev_name)return -EINVAL ;led_parse_fwnode_props (dev, fwnode, &props); +/* We want to label LEDs that can produce full range of colors +* as RGB, not multicolor */ +BUG_ON (props.color ==LED_COLOR_ID_MULTI ); +if (props.label) {/* * If init_data.devicename is NULL, then it indicates that Signature · 1 line
2.41.0
Quoted 5 lines · Mark Rutland
Quoted 4 lines · Marek Beh'un
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);the props variable is still initialized to zero.
Quoted 7 lines
Quoted 4 lines
Call the
BUG_ONonly after we parse fwnode into props.Fixes 77dce3a22e89 ("leds: disallow /sys/class/leds/*:multi:* for now")Signed-off-by Marek Beh'un <[email protected]>I've just discovered this has broken boot on my Libre Computer AML-A311D-CC-V0.2, which was working just fine with Debian 12's stock kernel:
Sorry the the noise; I've just spotted this is fixed by:
https://lore.kernel.org/linux-leds/[email protected]/
... and I look forward to that hitting mainline :)
Mark.
Lee Jones
maintainerQuoted 11 lines · Mark Rutland
Quoted 10 lines · Mark Rutland
Quoted 8 lines · Marek Beh'un
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);the props variable is still initialized to zero.
Call the
BUG_ONonly after we parse fwnode into props.Fixes 77dce3a22e89 ("leds: disallow /sys/class/leds/*:multi:* for now")Signed-off-by Marek Beh'un <[email protected]>
Quoted 8 lines
Quoted 2 lines
I've just discovered this has broken boot on my Libre Computer AML-A311D-CC-V0.2, which was working just fine with Debian 12's stock kernel:
Sorry the the noise; I've just spotted this is fixed by:
https://lore.kernel.org/linux-leds/[email protected]/
... and I look forward to that hitting mainline :)
Your long agonising wait is over:
commit 9dc1664fab2246bc2c3e9bf2cf21518a857f9b5b
Author: Marek Behún <[email protected]> Date: Mon Sep 18 16:07:24 2023 +0200
leds: Drop BUG_ON check for LED_COLOR_ID_MULTI
Commit c3f853184bed ("leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false") fixed a no-op BUG_ON. This turned out to cause a regression, since some in-tree device-tree files already use LED_COLOR_ID_MULTI.
Drop the BUG_ON altogether.
--- drivers/leds/led-core.c | 4 ---- 1 file changed, 4 deletions(-)
Signature · 1 line
Lee Jones [李琼斯]
Quoted 42 lines · Lee Jones
Quoted 11 lines · Mark Rutland
Quoted 10 lines · Mark Rutland
Quoted 8 lines · Marek Beh'un
At the time we call
BUG_ON(props.color == LED_COLOR_ID_MULTI);the props variable is still initialized to zero.
Call the
BUG_ONonly after we parse fwnode into props.Fixes 77dce3a22e89 ("leds: disallow /sys/class/leds/*:multi:* for now")Signed-off-by Marek Beh'un <[email protected]>Quoted 8 lines
Quoted 2 lines
I've just discovered this has broken boot on my Libre Computer AML-A311D-CC-V0.2, which was working just fine with Debian 12's stock kernel:
Sorry the the noise; I've just spotted this is fixed by:
https://lore.kernel.org/linux-leds/[email protected]/
... and I look forward to that hitting mainline :)
Your long agonising wait is over:
commit
9dc1664fab2246bc2c3e9bf2cf21518a857f9b5bAuthor: Marek Beh'un <[email protected]> Date: Mon Sep 18 16:07:24 2023 +0200leds: Drop
BUG_ONcheck forLED_COLOR_ID_MULTICommit
c3f853184bed("leds: FixBUG_ONcheck forLED_COLOR_ID_MULTIthat is always false") fixed a no-opBUG_ON. This turned out to cause a regression, since some in-tree device-tree files already useLED_COLOR_ID_MULTI.Drop the
BUG_ONaltogether.Fixes c3f853184bed ("leds: Fix BUG_ON check for LED_COLOR_ID_MULTI that is always false")Reported-by Da Xue <[email protected]>Signed-off-by Marek Beh'un <[email protected]>Signed-off-by Lee Jones <[email protected]>
Quoted 3 lines
--- drivers/leds/led-core.c | 4 ---- 1 file changed, 4 deletions(-)
Ah, doubly sorry then! I must've messed up when checking the logs for drivers/leds/...
Thanks, regardless!
Mark.