kernel/drivers/leds/led-class.c, branch linux-5.1.y

leds: class: ensure workqueue is initialized before setting brightness

2018-05-24T20:08:26Z

An application can try to set brightness before all the initialization is done, in particular before the workqueue is initialized with the call to led_init_core(). Here's a WARNING easy to trigger: [ 36.780813] WARNING: CPU: 3 PID: 1411 at ../kernel/workqueue.c:1444 __queue_work+0x37b/0x420 [ 36.780815] Modules linked in: ... [ 36.780868] CPU: 3 PID: 1411 Comm: systemd-backlig Not tainted 4.16.9-1-default #1 openSUSE Tumbleweed (unreleased) [ 36.780868] Hardware name: Dell Inc. Precision 5510/0N8J4R, BIOS 1.6.1 12/11/2017 [ 36.780870] RIP: 0010:__queue_work+0x37b/0x420 [ 36.780871] RSP: 0018:ffffaced048b7d78 EFLAGS: 00010086 [ 36.780873] RAX: 0000000000000000 RBX: ffffffffb3f01440 RCX: 0000000000000000 [ 36.780873] RDX: ffffffffc05a90d8 RSI: 0000000000000000 RDI: ffff8eac7dce2700 [ 36.780874] RBP: ffff8ea547c16400 R08: ffff8ea547800000 R09: ffff8eac7dc22700 [ 36.780875] R10: 0000000000000000 R11: 0000000000000040 R12: 0000000000000003 [ 36.780876] R13: 0000000000000200 R14: ffffffffc05a90d0 R15: ffff8eac7dce8600 [ 36.780877] FS: 00007f871e61cf40(0000) GS:ffff8eac7dcc0000(0000) knlGS:0000000000000000 [ 36.780878] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 36.780879] CR2: 000055c91115e308 CR3: 0000000883ee0005 CR4: 00000000003606e0 [ 36.780880] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 36.780880] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 36.780881] Call Trace: [ 36.780886] queue_work_on+0x81/0x90 [ 36.780889] brightness_store+0x5d/0x90 [ 36.780892] kernfs_fop_write+0x105/0x180 [ 36.780894] __vfs_write+0x26/0x150 [ 36.780897] ? common_file_perm+0x51/0x150 [ 36.780900] ? security_file_permission+0x3c/0xb0 [ 36.780901] vfs_write+0xad/0x1a0 [ 36.780903] SyS_write+0x42/0x90 [ 36.780906] do_syscall_64+0x76/0x140 [ 36.780908] entry_SYSCALL_64_after_hwframe+0x42/0xb7 [ 36.780910] RIP: 0033:0x7f871dd04c94 [ 36.780910] RSP: 002b:00007ffeb3a57d38 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 [ 36.780912] RAX: ffffffffffffffda RBX: 000055c91115c810 RCX: 00007f871dd04c94 [ 36.780912] RDX: 0000000000000001 RSI: 000055c91115c810 RDI: 0000000000000004 [ 36.780913] RBP: 00007ffeb3a57e10 R08: 0000000000000003 R09: 0000000000000000 [ 36.780914] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001 [ 36.780914] R13: 000055c911158f30 R14: 000055c90f3a9a4e R15: 0000000000000004 [ 36.780917] Code: 74 18 e8 49 80 00 00 48 85 c0 74 0e 48 8b 40 20 48 3b 68 08 0f 84 c2 fc ff ff 0f 0b 48 83 c4 10 5b 5d 41 5c 41 5d 41 5e 41 5f c3 <0f> 0b e9 82 fd ff ff 83 cd 02 49 8d 57 60 e9 69 fd ff ff 80 3d [ 36.780942] ---[ end trace 1fce4edad54c4017 ]--- This patch initializes and acquires the led_access mutex early in the of_led_classdev_register function, so that any application trying to write to sysfs to set brightness will block until initialization ends. Signed-off-by: Luis Henriques Acked-by: Pavel Machek Signed-off-by: Jacek Anaszewski

leds: core: add OF variants of LED registering functions

2017-03-08T20:10:01Z

These new functions allow passing an additional device_node argument that will be internally set for created LED device. Thanks to this LED core code and triggers will be able to access DT node for reading extra info. The easiest solution for achieving this was reworking old functions to more generic ones & adding simple defines for API compatibility. Signed-off-by: Rafał Miłecki Acked-by: Pavel Machek Signed-off-by: Jacek Anaszewski

leds: class: Add new optional brightness_hw_changed attribute

2017-01-29T18:59:42Z

Some LEDs may have their brightness level changed autonomously (outside of kernel control) by hardware / firmware. This commit adds support for an optional brightness_hw_changed attribute to signal such changes to userspace (if a driver can detect them): What: /sys/class/leds//brightness_hw_changed Date: January 2017 KernelVersion: 4.11 Description: Last hardware set brightness level for this LED. Some LEDs may be changed autonomously by hardware/firmware. Only LEDs where this happens and the driver can detect this, will have this file. This file supports poll() to detect when the hardware changes the brightness. Reading this file will return the last brightness level set by the hardware, this may be different from the current brightness. Drivers which want to support this, simply add LED_BRIGHT_HW_CHANGED to their flags field and call led_classdev_notify_brightness_hw_changed() with the hardware set brightness when they detect a hardware / firmware triggered brightness change. Signed-off-by: Hans de Goede Acked-by: Pavel Machek Signed-off-by: Jacek Anaszewski

led: core: Use atomic bit-field for the blink-flags

2016-11-22T11:07:05Z

All the LED_BLINK* flags are accessed read-modify-write from e.g. led_set_brightness and led_blink_set_oneshot while both set_brightness_work and the blink_timer may be running. If these race then the modify step done by one of them may be lost, switch the LED_BLINK* flags to a new atomic work_flags bit-field to avoid this race. Signed-off-by: Hans de Goede Signed-off-by: Jacek Anaszewski

leds: Use macro for max device node name size

2016-11-22T11:07:02Z

Use a macro instead of hard-coding the max device node name size. The uleds driver introduced a macro for this value, so using it. Signed-off-by: David Lechner Signed-off-by: Jacek Anaszewski

leds: core: avoid error message when a USB LED device is unplugged

2016-03-14T08:22:20Z

When a USB LED device is unplugged the remove call chain calls led_classdev_unregister which tries to switch the LED off. As the device has been removed already this results in a ENODEV error message in dmesg. Avoid this error message by ignoring ENODEV in calls from led_classdev_unregister if the LED device is flagged as pluggable. Therefore a new flag LED_HW_PLUGGABLE was introduced which should be set by all LED drivers handling pluggable LED devices (mainly USB LED devices). Signed-off-by: Heiner Kallweit Signed-off-by: Jacek Anaszewski

leds: turn off the LED and wait for completion on unregistering LED class device

2016-01-04T08:57:37Z

Workqueue, 'set_brightness_work' is used for scheduling brightness control. This workqueue is canceled when the LED class device is unregistered. Currently, LED subsystem handles like below. cancel_work_sync(&led_cdev->set_brightness_work) led_set_brightness(led_cdev, LED_OFF) However, this could be a problem. Workqueue is going to be canceled but LED device needs to be off. The worst case is null pointer access due to scheduling a workqueue. LED module is loaded. LED driver private data is allocated by using devm_zalloc(). LED module is unloaded. led_classdev_unregister() is called. cancel_work_sync() led_set_brightness(led_cdev, LED_OFF) schedule_work() if LED driver uses brightness_set_blocking() In the meantime, driver private data will be freed. ..scheduling.. brightness_set_blocking() callback is invoked. For the brightness control, LED driver tries to access private data but resource is removed! To avoid this problem, LED subsystem should turn off the brightness first and wait for completion. led_set_brightness(led_cdev, LED_OFF) flush_work(&led_cdev->set_brightness_work) It guarantees that LED driver turns off the brightness prior to resource management. Cc: linux-leds@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Milo Kim Signed-off-by: Jacek Anaszewski

leds: core: Drivers shouldn't enforce SYNC/ASYNC brightness setting

2016-01-04T08:57:31Z

This patch removes SET_BRIGHTNESS_ASYNC and SET_BRIGHTNESS_SYNC flags. led_set_brightness() now calls led_set_brightness_nosleep() instead of choosing between sync and async op basing on the flags defined by the driver. From now on, if a user wants to make sure that brightness will be set synchronously, they have to use led_set_brightness_sync() API. It is now being made publicly available since it has become apparent that it is a caller who should decide whether brightness is to be set in a synchronous or an asynchronous way. Signed-off-by: Jacek Anaszewski Acked-by: Sakari Ailus

leds: core: Add led_set_brightness_nosleep{nopm} functions

2016-01-04T08:57:30Z

This patch adds led_set_brightness_nosleep() and led_set_brightness_nopm() functions, that guarantee setting LED brightness in a non-blocking way. The latter is used from pm_ops context and doesn't modify the brightness cached in the struct led_classdev. Its execution always ends up with a call to brightness setting op - either directly or through a set_brightness_work, regardless of LED_SUSPENDED flag state. The patch also replaces led_set_brightness_async() with led_set_brightness_nosleep() in all places where the most vital was setting brightness in a non sleeping way but not necessarily asynchronously, which is not needed for non-blocking drivers. Signed-off-by: Jacek Anaszewski Acked-by: Sakari Ailus

leds: core: Move LED core callbacks out of led-class.c

2015-11-03T07:59:22Z

Since the API for controlling LED brightness and blinking is defined in the LED core, move the related timer and work callbacks to the led-core.c, and initialize them through a new led_core_init API. Signed-off-by: Jacek Anaszewski Acked-by: Andrew Lunn Acked-by: Pavel Machek