All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
@ 2026-11-20  7:56 Zheng Shuhao
  2026-11-20  9:34 ` sashiko-bot
                   ` (3 more replies)
  0 siblings, 4 replies; 13+ messages in thread
From: Zheng Shuhao @ 2026-11-20  7:56 UTC (permalink / raw)
  To: Jiri Kosina
  Cc: Jonathan Cameron, Srinivas Pandruvada, Benjamin Tissoires,
	linux-input, linux-iio, linux-kernel,
	syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f

syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
sensor hub removal.

sensor_hub_remove() currently holds sensor_data->lock while calling
device_for_each_child(). The child iteration enters the device core and may
acquire the klist lock. Since sensor_data->lock is also acquired with
interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
lock-order inversion.

The callback only needs sensor_data->lock while checking pending.status and
completing pending.ready. Pass sensor_data to the callback and take the lock
only around that short critical section instead of holding it across
device_for_each_child().

This preserves synchronization of pending.status while avoiding the lock-order
dependency between sensor_data->lock and the device core's klist lock.

The syzbot report is available at:
https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f

Testing:
- Reproduced the target lockdep warning with the unmodified kernel in 3/3
  two-CPU QEMU runs.
- The warning was absent with this patch in 3/3 equivalent runs.
- Ran the HID selftests against both baseline and patched kernels:
  hid_bpf: 45 passed, 1 skipped, 0 failed
  hidraw: 27 passed, 0 failed
  hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
  hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
- hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
  kernels because no executable test cases were collected in this minimal
  QEMU environment.
- No KASAN report, use-after-free, Oops, BUG, panic, or general protection
  fault was observed.
- checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.

Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
---
 drivers/hid/hid-sensor-hub.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-hub.c
index 6470a290ebfc5..62799c26e6ec3 100644
--- a/drivers/hid/hid-sensor-hub.c
+++ b/drivers/hid/hid-sensor-hub.c
@@ -804,9 +804,13 @@ static int sensor_hub_probe(struct hid_device *hdev,
 static int sensor_hub_finalize_pending_fn(struct device *dev, void *data)
 {
 	struct hid_sensor_hub_device *hsdev = dev->platform_data;
+	struct sensor_hub_data *sensor_data = data;
+	unsigned long flags;
 
+	spin_lock_irqsave(&sensor_data->lock, flags);
 	if (hsdev->pending.status)
 		complete(&hsdev->pending.ready);
+	spin_unlock_irqrestore(&sensor_data->lock, flags);
 
 	return 0;
 }
@@ -814,16 +818,13 @@ static int sensor_hub_finalize_pending_fn(struct device *dev, void *data)
 static void sensor_hub_remove(struct hid_device *hdev)
 {
 	struct sensor_hub_data *data = hid_get_drvdata(hdev);
-	unsigned long flags;
 
 	hid_dbg(hdev, " hardware removed\n");
 	hid_hw_close(hdev);
 	hid_hw_stop(hdev);
 
-	spin_lock_irqsave(&data->lock, flags);
-	device_for_each_child(&hdev->dev, NULL,
+	device_for_each_child(&hdev->dev, data,
 			      sensor_hub_finalize_pending_fn);
-	spin_unlock_irqrestore(&data->lock, flags);
 
 	mfd_remove_devices(&hdev->dev);
 	mutex_destroy(&data->mutex);


^ permalink raw reply related	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20  7:56 [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration Zheng Shuhao
@ 2026-11-20  9:34 ` sashiko-bot
  2026-11-20 10:50 ` Joshua Crofts
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 13+ messages in thread
From: sashiko-bot @ 2026-11-20  9:34 UTC (permalink / raw)
  To: Zheng Shuhao; +Cc: linux-input, dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Low] HID: sensor-hub: missing Fixes trailer for commit 8a5b38c3fd70
It also noticed 1 related pre-existing issue this patch does not
introduce:
- [High] HID: sensor-hub: NULL pointer dereference in
  sensor_hub_finalize_pending_fn() [1]
--

commit 243a16dd0439f8612b7bbc316ea2122b572ceb3b
Author: Zheng Shuhao <viennaoffh@gmail.com>

> HID: sensor-hub: Avoid holding spinlock across child iteration
>
> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
> sensor hub removal.
>
> sensor_hub_remove() currently holds sensor_data->lock while calling
> device_for_each_child(). The child iteration enters the device core and may
> acquire the klist lock. Since sensor_data->lock is also acquired with
> interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
> lock-order inversion.
>
> The callback only needs sensor_data->lock while checking pending.status and
> completing pending.ready. Pass sensor_data to the callback and take the lock
> only around that short critical section instead of holding it across
> device_for_each_child().
>
> This preserves synchronization of pending.status while avoiding the lock-order
> dependency between sensor_data->lock and the device core's klist lock.
>
> The syzbot report is available at:
> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>
> Testing:
> - Reproduced the target lockdep warning with the unmodified kernel in 3/3
>   two-CPU QEMU runs.
> - The warning was absent with this patch in 3/3 equivalent runs.
> - Ran the HID selftests against both baseline and patched kernels:
>   hid_bpf: 45 passed, 1 skipped, 0 failed
>   hidraw: 27 passed, 0 failed
>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>   kernels because no executable test cases were collected in this minimal
>   QEMU environment.
> - No KASAN report, use-after-free, Oops, BUG, panic, or general protection
>   fault was observed.
> - checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.
>
> Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>

[Severity: Low]
This fixes the device_for_each_child() call under data->lock added in commit
8a5b38c3fd70 ("HID: hid-sensor-hub: don't use stale platform-data on
remove").  Should this include:

  Fixes: 8a5b38c3fd70 ("HID: hid-sensor-hub: don't use stale platform-data on remove")

> diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-hub.c
> index 028e70ec7bace..81015f5b82359 100644
> --- a/drivers/hid/hid-sensor-hub.c
> +++ b/drivers/hid/hid-sensor-hub.c
> @@ -836,9 +836,13 @@ static int sensor_hub_probe(struct hid_device *hdev,
>  static int sensor_hub_finalize_pending_fn(struct device *dev, void *data)
>  {
>  	struct hid_sensor_hub_device *hsdev = dev->platform_data;
> +	struct sensor_hub_data *sensor_data = data;
> +	unsigned long flags;
>  
> +	spin_lock_irqsave(&sensor_data->lock, flags);
>  	if (hsdev->pending.status)
>  		complete(&hsdev->pending.ready);
> +	spin_unlock_irqrestore(&sensor_data->lock, flags);

[Severity: High]
This problem wasn't introduced by this patch, but can hsdev be NULL here and
cause a NULL pointer dereference [1]?

When CONFIG_HID_BATTERY_STRENGTH is enabled and a composite HID sensor hub
device also exposes a battery strength usage, hid_hw_start() ->
hidinput_setup_battery() registers a devm-managed power_supply child device
under &hdev->dev with dev->platform_data set to NULL.  On teardown,
sensor_hub_remove() runs before hid_device_remove() releases the devres
group, and hid_hw_stop() only removes input_dev children, so the
power_supply child is still attached when device_for_each_child() calls
sensor_hub_finalize_pending_fn().

[1] https://sashiko.dev/#/bug/linux-23655093-642a-447c-b416-776e47c98d55

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/1795161418.hid-sensor-hub@viennaoffh.gmail.com?part=1

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20  7:56 [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration Zheng Shuhao
  2026-11-20  9:34 ` sashiko-bot
@ 2026-11-20 10:50 ` Joshua Crofts
  2026-11-20 10:50   ` syzbot
  2026-11-20 11:09   ` Joshua Crofts
  2026-11-20 12:14 ` Srinivas Pandruvada
  2026-11-20 13:50 ` [PATCH v2] " Zheng Shuhao
  3 siblings, 2 replies; 13+ messages in thread
From: Joshua Crofts @ 2026-11-20 10:50 UTC (permalink / raw)
  To: Zheng Shuhao, Jiri Kosina
  Cc: Jonathan Cameron, Srinivas Pandruvada, Benjamin Tissoires,
	linux-input, linux-iio, linux-kernel,
	syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f

On 20/11/2026 08:56, Zheng Shuhao wrote:
> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
> sensor hub removal.
> 
> sensor_hub_remove() currently holds sensor_data->lock while calling
> device_for_each_child(). The child iteration enters the device core and may
> acquire the klist lock. Since sensor_data->lock is also acquired with
> interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
> lock-order inversion.
> 
> The callback only needs sensor_data->lock while checking pending.status and
> completing pending.ready. Pass sensor_data to the callback and take the lock
> only around that short critical section instead of holding it across
> device_for_each_child().
> 
> This preserves synchronization of pending.status while avoiding the lock-order
> dependency between sensor_data->lock and the device core's klist lock.
> 
> The syzbot report is available at:
> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
> 
> Testing:
> - Reproduced the target lockdep warning with the unmodified kernel in 3/3
>   two-CPU QEMU runs.
> - The warning was absent with this patch in 3/3 equivalent runs.
> - Ran the HID selftests against both baseline and patched kernels:
>   hid_bpf: 45 passed, 1 skipped, 0 failed
>   hidraw: 27 passed, 0 failed
>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>   kernels because no executable test cases were collected in this minimal
>   QEMU environment.
> - No KASAN report, use-after-free, Oops, BUG, panic, or general protection
>   fault was observed.
> - checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.
> 
> Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
You cannot add a Tested-by tag with your name.  This is reserved for when
someone independent of you tests your patch. Looking at the syzbot report
there is a reproducer, so syzbot should test it. You can make syzbot test
a patch by sending the following (note, this will trigger a test):

#syz test

-- 
Kind regards,
Joshua Crofts


^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 10:50 ` Joshua Crofts
@ 2026-11-20 10:50   ` syzbot
  2026-11-20 11:09   ` Joshua Crofts
  1 sibling, 0 replies; 13+ messages in thread
From: syzbot @ 2026-11-20 10:50 UTC (permalink / raw)
  To: crofts
  Cc: bentiss, crofts, jic23, jikos, linux-iio, linux-input,
	linux-kernel, srinivas.pandruvada, viennaoffh

> On 20/11/2026 08:56, Zheng Shuhao wrote:
>> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
>> sensor hub removal.
>> 
>> sensor_hub_remove() currently holds sensor_data->lock while calling
>> device_for_each_child(). The child iteration enters the device core and may
>> acquire the klist lock. Since sensor_data->lock is also acquired with
>> interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
>> lock-order inversion.
>> 
>> The callback only needs sensor_data->lock while checking pending.status and
>> completing pending.ready. Pass sensor_data to the callback and take the lock
>> only around that short critical section instead of holding it across
>> device_for_each_child().
>> 
>> This preserves synchronization of pending.status while avoiding the lock-order
>> dependency between sensor_data->lock and the device core's klist lock.
>> 
>> The syzbot report is available at:
>> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>> 
>> Testing:
>> - Reproduced the target lockdep warning with the unmodified kernel in 3/3
>>   two-CPU QEMU runs.
>> - The warning was absent with this patch in 3/3 equivalent runs.
>> - Ran the HID selftests against both baseline and patched kernels:
>>   hid_bpf: 45 passed, 1 skipped, 0 failed
>>   hidraw: 27 passed, 0 failed
>>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
>> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>>   kernels because no executable test cases were collected in this minimal
>>   QEMU environment.
>> - No KASAN report, use-after-free, Oops, BUG, panic, or general protection
>>   fault was observed.
>> - checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.
>> 
>> Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
>> Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
>> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
> You cannot add a Tested-by tag with your name.  This is reserved for when
> someone independent of you tests your patch. Looking at the syzbot report
> there is a reproducer, so syzbot should test it. You can make syzbot test
> a patch by sending the following (note, this will trigger a test):
>
> #syz test

I see the command but can't find the corresponding bug.
The email is sent to  syzbot+HASH@syzkaller.appspotmail.com address
but the HASH does not correspond to any known bug.
Please double check the address.

>
> -- 
> Kind regards,
> Joshua Crofts
>

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 10:50 ` Joshua Crofts
  2026-11-20 10:50   ` syzbot
@ 2026-11-20 11:09   ` Joshua Crofts
  2026-11-20 11:09     ` syzbot
  1 sibling, 1 reply; 13+ messages in thread
From: Joshua Crofts @ 2026-11-20 11:09 UTC (permalink / raw)
  To: Zheng Shuhao, Jiri Kosina
  Cc: Jonathan Cameron, Srinivas Pandruvada, Benjamin Tissoires,
	linux-input, linux-iio, linux-kernel, syzbot+fb951970a334aaff00ba



On 20/11/2026 11:50, Joshua Crofts wrote:
> On 20/11/2026 08:56, Zheng Shuhao wrote:
>> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
>> sensor hub removal.
>> 
>> sensor_hub_remove() currently holds sensor_data->lock while calling
>> device_for_each_child(). The child iteration enters the device core and may
>> acquire the klist lock. Since sensor_data->lock is also acquired with
>> interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
>> lock-order inversion.
>> 
>> The callback only needs sensor_data->lock while checking pending.status and
>> completing pending.ready. Pass sensor_data to the callback and take the lock
>> only around that short critical section instead of holding it across
>> device_for_each_child().
>> 
>> This preserves synchronization of pending.status while avoiding the lock-order
>> dependency between sensor_data->lock and the device core's klist lock.
>> 
>> The syzbot report is available at:
>> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>> 
>> Testing:
>> - Reproduced the target lockdep warning with the unmodified kernel in 3/3
>>   two-CPU QEMU runs.
>> - The warning was absent with this patch in 3/3 equivalent runs.
>> - Ran the HID selftests against both baseline and patched kernels:
>>   hid_bpf: 45 passed, 1 skipped, 0 failed
>>   hidraw: 27 passed, 0 failed
>>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
>> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>>   kernels because no executable test cases were collected in this minimal
>>   QEMU environment.
>> - No KASAN report, use-after-free, Oops, BUG, panic, or general protection
>>   fault was observed.
>> - checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.
>> 
>> Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
>> Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
>> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
> You cannot add a Tested-by tag with your name.  This is reserved for when
> someone independent of you tests your patch. Looking at the syzbot report
> there is a reproducer, so syzbot should test it. You can make syzbot test
> a patch by sending the following (note, this will trigger a test):

Ah, did some digging and the syzbot email is incorrect, it should be:

syzbot+fb951970a334aaff00ba@syzkaller.appspotmail.com

Cc'd the right syzbot and triggering the test.

#syz test

-- 
Kind regards,
Joshua Crofts


^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 11:09   ` Joshua Crofts
@ 2026-11-20 11:09     ` syzbot
  0 siblings, 0 replies; 13+ messages in thread
From: syzbot @ 2026-11-20 11:09 UTC (permalink / raw)
  To: crofts
  Cc: bentiss, crofts, jic23, jikos, linux-iio, linux-input,
	linux-kernel, srinivas.pandruvada, viennaoffh, syzkaller-bugs

>
>
> On 20/11/2026 11:50, Joshua Crofts wrote:
>> On 20/11/2026 08:56, Zheng Shuhao wrote:
>>> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning during
>>> sensor hub removal.
>>> 
>>> sensor_hub_remove() currently holds sensor_data->lock while calling
>>> device_for_each_child(). The child iteration enters the device core and may
>>> acquire the klist lock. Since sensor_data->lock is also acquired with
>>> interrupts disabled by sensor_hub_raw_event(), lockdep reports a possible
>>> lock-order inversion.
>>> 
>>> The callback only needs sensor_data->lock while checking pending.status and
>>> completing pending.ready. Pass sensor_data to the callback and take the lock
>>> only around that short critical section instead of holding it across
>>> device_for_each_child().
>>> 
>>> This preserves synchronization of pending.status while avoiding the lock-order
>>> dependency between sensor_data->lock and the device core's klist lock.
>>> 
>>> The syzbot report is available at:
>>> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>>> 
>>> Testing:
>>> - Reproduced the target lockdep warning with the unmodified kernel in 3/3
>>>   two-CPU QEMU runs.
>>> - The warning was absent with this patch in 3/3 equivalent runs.
>>> - Ran the HID selftests against both baseline and patched kernels:
>>>   hid_bpf: 45 passed, 1 skipped, 0 failed
>>>   hidraw: 27 passed, 0 failed
>>>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>>>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
>>> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>>>   kernels because no executable test cases were collected in this minimal
>>>   QEMU environment.
>>> - No KASAN report, use-after-free, Oops, BUG, panic, or general protection
>>>   fault was observed.
>>> - checkpatch.pl --strict --no-signoff reports 0 errors and 0 warnings.
>>> 
>>> Reported-by: syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
>>> Closes: https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
>>> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
>>> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
>> You cannot add a Tested-by tag with your name.  This is reserved for when
>> someone independent of you tests your patch. Looking at the syzbot report
>> there is a reproducer, so syzbot should test it. You can make syzbot test
>> a patch by sending the following (note, this will trigger a test):
>
> Ah, did some digging and the syzbot email is incorrect, it should be:
>
> syzbot+fb951970a334aaff00ba@syzkaller.appspotmail.com
>
> Cc'd the right syzbot and triggering the test.
>
> #syz test

Patch testing via #syz test is temporarily disabled.
See https://github.com/google/syzkaller/issues/8122 for details.

>
> -- 
> Kind regards,
> Joshua Crofts
>

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20  7:56 [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration Zheng Shuhao
  2026-11-20  9:34 ` sashiko-bot
  2026-11-20 10:50 ` Joshua Crofts
@ 2026-11-20 12:14 ` Srinivas Pandruvada
  2026-11-20 13:49   ` Zheng Shuhao
  2026-11-20 13:50 ` [PATCH v2] " Zheng Shuhao
  3 siblings, 1 reply; 13+ messages in thread
From: Srinivas Pandruvada @ 2026-11-20 12:14 UTC (permalink / raw)
  To: Zheng Shuhao, Jiri Kosina, ayaanmirzabaig85
  Cc: Jonathan Cameron, Benjamin Tissoires, linux-input, linux-iio,
	linux-kernel, syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f

On Fri, 2026-11-20 at 15:56 +0800, Zheng Shuhao wrote:
> syzbot reports a SOFTIRQ-safe -> SOFTIRQ-unsafe lock order warning
> during
> sensor hub removal.
> 
> sensor_hub_remove() currently holds sensor_data->lock while calling
> device_for_each_child(). The child iteration enters the device core
> and may
> acquire the klist lock. Since sensor_data->lock is also acquired with
> interrupts disabled by sensor_hub_raw_event(), lockdep reports a
> possible
> lock-order inversion.
> 
> The callback only needs sensor_data->lock while checking
> pending.status and
> completing pending.ready. Pass sensor_data to the callback and take
> the lock
> only around that short critical section instead of holding it across
> device_for_each_child().
> 
> This preserves synchronization of pending.status while avoiding the
> lock-order
> dependency between sensor_data->lock and the device core's klist
> lock.
> 
> The syzbot report is available at:
> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
> 

Ayaan posted a similar patch yesterday.

Thanks,
Srinivas


> Testing:
> - Reproduced the target lockdep warning with the unmodified kernel in
> 3/3
>   two-CPU QEMU runs.
> - The warning was absent with this patch in 3/3 equivalent runs.
> - Ran the HID selftests against both baseline and patched kernels:
>   hid_bpf: 45 passed, 1 skipped, 0 failed
>   hidraw: 27 passed, 0 failed
>   hid-core, hid-apple, hid-gamepad, hid-ite, hid-keyboard, hid-mouse,
>   hid-multitouch, hid-sony, and hid-tablet all passed in both runs.
> - hid-usb_crash.sh and hid-wacom.sh produced TAP 1..0 / rc=2 in both
>   kernels because no executable test cases were collected in this
> minimal
>   QEMU environment.
> - No KASAN report, use-after-free, Oops, BUG, panic, or general
> protection
>   fault was observed.
> - checkpatch.pl --strict --no-signoff reports 0 errors and 0
> warnings.
> 
> Reported-by:
> syzbot+c7c429f6851638ece81338f90c5b29ac9c0dea1f@syzkaller.appspotmail.com
> Closes:
> https://syzkaller.appspot.com/bug?id=c7c429f6851638ece81338f90c5b29ac9c0dea1f
> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
> Tested-by: Zheng Shuhao <viennaoffh@gmail.com>
> ---
>  drivers/hid/hid-sensor-hub.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-
> hub.c
> index 6470a290ebfc5..62799c26e6ec3 100644
> --- a/drivers/hid/hid-sensor-hub.c
> +++ b/drivers/hid/hid-sensor-hub.c
> @@ -804,9 +804,13 @@ static int sensor_hub_probe(struct hid_device
> *hdev,
>  static int sensor_hub_finalize_pending_fn(struct device *dev, void
> *data)
>  {
>  	struct hid_sensor_hub_device *hsdev = dev->platform_data;
> +	struct sensor_hub_data *sensor_data = data;
> +	unsigned long flags;
>  
> +	spin_lock_irqsave(&sensor_data->lock, flags);
>  	if (hsdev->pending.status)
>  		complete(&hsdev->pending.ready);
> +	spin_unlock_irqrestore(&sensor_data->lock, flags);
>  
>  	return 0;
>  }
> @@ -814,16 +818,13 @@ static int
> sensor_hub_finalize_pending_fn(struct device *dev, void *data)
>  static void sensor_hub_remove(struct hid_device *hdev)
>  {
>  	struct sensor_hub_data *data = hid_get_drvdata(hdev);
> -	unsigned long flags;
>  
>  	hid_dbg(hdev, " hardware removed\n");
>  	hid_hw_close(hdev);
>  	hid_hw_stop(hdev);
>  
> -	spin_lock_irqsave(&data->lock, flags);
> -	device_for_each_child(&hdev->dev, NULL,
> +	device_for_each_child(&hdev->dev, data,
>  			      sensor_hub_finalize_pending_fn);
> -	spin_unlock_irqrestore(&data->lock, flags);
>  
>  	mfd_remove_devices(&hdev->dev);
>  	mutex_destroy(&data->mutex);

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 12:14 ` Srinivas Pandruvada
@ 2026-11-20 13:49   ` Zheng Shuhao
  0 siblings, 0 replies; 13+ messages in thread
From: Zheng Shuhao @ 2026-11-20 13:49 UTC (permalink / raw)
  To: Srinivas Pandruvada, Jiri Kosina, Ayaan Mirza Baig
  Cc: Jonathan Cameron, Benjamin Tissoires, linux-input, linux-iio,
	linux-kernel

> Ayaan Mirza Baig posted a similar patch yesterday.

Sorry, I missed Ayaan's earlier patch. I've refined the fix to pass the
callback data and skip children without Sensor Hub platform_data, and am
sending v2 separately. The #syz test request was rejected because syzbot's
automated patch testing is temporarily unavailable.

Thanks,
Zheng

^ permalink raw reply	[flat|nested] 13+ messages in thread

* [PATCH v2] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20  7:56 [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration Zheng Shuhao
                   ` (2 preceding siblings ...)
  2026-11-20 12:14 ` Srinivas Pandruvada
@ 2026-11-20 13:50 ` Zheng Shuhao
  2026-11-20 15:12   ` Srinivas Pandruvada
  3 siblings, 1 reply; 13+ messages in thread
From: Zheng Shuhao @ 2026-11-20 13:50 UTC (permalink / raw)
  To: linux-input
  Cc: jikos, jic23, srinivas.pandruvada, bentiss, heiko, lee, linux-iio,
	linux-kernel, ayaanmirzabaig85, syzbot+fb951970a334aaff00ba

sensor_hub_remove() holds sensor_data->lock while walking the HID device's
children with device_for_each_child(). The child iteration takes the device
core's klist lock, while sensor_hub_raw_event() takes
sensor_data->lock with interrupts disabled. This creates a lock-order
inversion.

Pass sensor_data to the callback and take the lock only while examining
pending.status and completing pending.ready.

The callback also visits non-Sensor Hub children. When HID battery strength
support is enabled, the HID input layer may register a power-supply child
without platform_data. Skip children without Sensor Hub data before
dereferencing it.

Testing:
- With CONFIG_HID_BATTERY_STRENGTH=y, ran a UHID Sensor Hub/Joystick device
  exposing Battery Strength through 15 create/destroy cycles across three
  QEMU boots. The power-supply child was found each time; no Oops, BUG, or
  KASAN report occurred.
- In the same configuration, hid_bpf passed 45 tests with 1 skip, hidraw
  passed 27 tests, and hid-core, hid-apple, hid-gamepad, hid-ite,
  hid-keyboard, and hid-mouse passed.
- The full HID suite was not completed; it stopped during hid-multitouch.
- The #syz test request was rejected because syzbot's automated patch
  testing is temporarily unavailable.

Fixes: 8a5b38c3fd70 ("HID: hid-sensor-hub: don't use stale platform-data on remove")
Reported-by: syzbot+fb951970a334aaff00ba@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=fb951970a334aaff00ba
Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
---
Changes in v2:
- Pass sensor_data to the callback instead of NULL.
- Skip children without Sensor Hub platform_data; add a battery-child test.
- Add Fixes and correct the syzbot address; remove the self-added Tested-by.

 drivers/hid/hid-sensor-hub.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-hub.c
index 6470a290ebfc5..4c447f62480b5 100644
--- a/drivers/hid/hid-sensor-hub.c
+++ b/drivers/hid/hid-sensor-hub.c
@@ -804,9 +804,16 @@ static int sensor_hub_probe(struct hid_device *hdev,
 static int sensor_hub_finalize_pending_fn(struct device *dev, void *data)
 {
 	struct hid_sensor_hub_device *hsdev = dev->platform_data;
+	struct sensor_hub_data *sensor_data = data;
+	unsigned long flags;
+
+	if (!hsdev)
+		return 0;
 
+	spin_lock_irqsave(&sensor_data->lock, flags);
 	if (hsdev->pending.status)
 		complete(&hsdev->pending.ready);
+	spin_unlock_irqrestore(&sensor_data->lock, flags);
 
 	return 0;
 }
@@ -814,16 +821,13 @@ static int sensor_hub_finalize_pending_fn(struct device *dev, void *data)
 static void sensor_hub_remove(struct hid_device *hdev)
 {
 	struct sensor_hub_data *data = hid_get_drvdata(hdev);
-	unsigned long flags;
 
 	hid_dbg(hdev, " hardware removed\n");
 	hid_hw_close(hdev);
 	hid_hw_stop(hdev);
 
-	spin_lock_irqsave(&data->lock, flags);
-	device_for_each_child(&hdev->dev, NULL,
+	device_for_each_child(&hdev->dev, data,
 			      sensor_hub_finalize_pending_fn);
-	spin_unlock_irqrestore(&data->lock, flags);
 
 	mfd_remove_devices(&hdev->dev);
 	mutex_destroy(&data->mutex);
-- 

^ permalink raw reply related	[flat|nested] 13+ messages in thread

* Re: [PATCH v2] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 13:50 ` [PATCH v2] " Zheng Shuhao
@ 2026-11-20 15:12   ` Srinivas Pandruvada
  2026-11-20 15:30     ` Srinivas Pandruvada
  0 siblings, 1 reply; 13+ messages in thread
From: Srinivas Pandruvada @ 2026-11-20 15:12 UTC (permalink / raw)
  To: Zheng Shuhao, linux-input, Zhang Lixu
  Cc: jikos, jic23, bentiss, heiko, lee, linux-iio, linux-kernel,
	ayaanmirzabaig85, syzbot+fb951970a334aaff00ba

+Lixu

On Fri, 2026-11-20 at 21:50 +0800, Zheng Shuhao wrote:
> sensor_hub_remove() holds sensor_data->lock while walking the HID
> device's
> children with device_for_each_child(). The child iteration takes the
> device
> core's klist lock, while sensor_hub_raw_event() takes
> sensor_data->lock with interrupts disabled. This creates a lock-order
> inversion.
> 
> Pass sensor_data to the callback and take the lock only while
> examining
> pending.status and completing pending.ready.
> 
> The callback also visits non-Sensor Hub children. When HID battery
> strength
> support is enabled, the HID input layer may register a power-supply
> child
> without platform_data. Skip children without Sensor Hub data before
> dereferencing it.
> 
> Testing:
> - With CONFIG_HID_BATTERY_STRENGTH=y, ran a UHID Sensor Hub/Joystick
> device
>   exposing Battery Strength through 15 create/destroy cycles across
> three
>   QEMU boots. The power-supply child was found each time; no Oops,
> BUG, or
>   KASAN report occurred.
> - In the same configuration, hid_bpf passed 45 tests with 1 skip,
> hidraw
>   passed 27 tests, and hid-core, hid-apple, hid-gamepad, hid-ite,
>   hid-keyboard, and hid-mouse passed.
> - The full HID suite was not completed; it stopped during hid-
> multitouch.
> - The #syz test request was rejected because syzbot's automated patch
>   testing is temporarily unavailable.

None of the above tested devices are HID SENSOR HUB related, so need
some tests.
Hi Lixu,

Please give a try.

Thanks,
Srinivas

> 
> Fixes: 8a5b38c3fd70 ("HID: hid-sensor-hub: don't use stale platform-
> data on remove")
> Reported-by: syzbot+fb951970a334aaff00ba@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=fb951970a334aaff00ba
> Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
> ---
> Changes in v2:
> - Pass sensor_data to the callback instead of NULL.
> - Skip children without Sensor Hub platform_data; add a battery-child
> test.
> - Add Fixes and correct the syzbot address; remove the self-added
> Tested-by.
> 
>  drivers/hid/hid-sensor-hub.c | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-
> hub.c
> index 6470a290ebfc5..4c447f62480b5 100644
> --- a/drivers/hid/hid-sensor-hub.c
> +++ b/drivers/hid/hid-sensor-hub.c
> @@ -804,9 +804,16 @@ static int sensor_hub_probe(struct hid_device
> *hdev,
>  static int sensor_hub_finalize_pending_fn(struct device *dev, void
> *data)
>  {
>  	struct hid_sensor_hub_device *hsdev = dev->platform_data;
> +	struct sensor_hub_data *sensor_data = data;
> +	unsigned long flags;
> +
> +	if (!hsdev)
> +		return 0;
>  
> +	spin_lock_irqsave(&sensor_data->lock, flags);
>  	if (hsdev->pending.status)
>  		complete(&hsdev->pending.ready);
> +	spin_unlock_irqrestore(&sensor_data->lock, flags);
>  
>  	return 0;
>  }
> @@ -814,16 +821,13 @@ static int
> sensor_hub_finalize_pending_fn(struct device *dev, void *data)
>  static void sensor_hub_remove(struct hid_device *hdev)
>  {
>  	struct sensor_hub_data *data = hid_get_drvdata(hdev);
> -	unsigned long flags;
>  
>  	hid_dbg(hdev, " hardware removed\n");
>  	hid_hw_close(hdev);
>  	hid_hw_stop(hdev);
>  
> -	spin_lock_irqsave(&data->lock, flags);
> -	device_for_each_child(&hdev->dev, NULL,
> +	device_for_each_child(&hdev->dev, data,
>  			      sensor_hub_finalize_pending_fn);
> -	spin_unlock_irqrestore(&data->lock, flags);
>  
>  	mfd_remove_devices(&hdev->dev);
>  	mutex_destroy(&data->mutex);

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH v2] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 15:12   ` Srinivas Pandruvada
@ 2026-11-20 15:30     ` Srinivas Pandruvada
  2026-11-20 17:05       ` Zheng Shuhao
  0 siblings, 1 reply; 13+ messages in thread
From: Srinivas Pandruvada @ 2026-11-20 15:30 UTC (permalink / raw)
  To: Zheng Shuhao, linux-input, Zhang Lixu
  Cc: jikos, jic23, bentiss, heiko, lee, linux-iio, linux-kernel,
	ayaanmirzabaig85, syzbot+fb951970a334aaff00ba

On Fri, 2026-11-20 at 07:12 -0800, Srinivas Pandruvada wrote:
> +Lixu
> 
> On Fri, 2026-11-20 at 21:50 +0800, Zheng Shuhao wrote:
> > sensor_hub_remove() holds sensor_data->lock while walking the HID
> > device's
> > children with device_for_each_child(). The child iteration takes
> > the
> > device
> > core's klist lock, while sensor_hub_raw_event() takes
> > sensor_data->lock with interrupts disabled. This creates a lock-
> > order
> > inversion.
> > 

In reality this can't happen as the
hid_hw_close() and hid_hw_stop() is called before, so there can't be
any raw_event.

> > Pass sensor_data to the callback and take the lock only while
> > examining
> > pending.status and completing pending.ready.
> > 
> > The callback also visits non-Sensor Hub children. When HID battery
> > strength
> > support is enabled, the HID input layer may register a power-supply
> > child
> > without platform_data. Skip children without Sensor Hub data before
> > dereferencing it.
> > 

How come this is possible? The HID sensor hub is only deleting its own
children. Can you give an example here?

Thanks,
Srinivas


> > Testing:
> > - With CONFIG_HID_BATTERY_STRENGTH=y, ran a UHID Sensor
> > Hub/Joystick
> > device
> >   exposing Battery Strength through 15 create/destroy cycles across
> > three
> >   QEMU boots. The power-supply child was found each time; no Oops,
> > BUG, or
> >   KASAN report occurred.
> > - In the same configuration, hid_bpf passed 45 tests with 1 skip,
> > hidraw
> >   passed 27 tests, and hid-core, hid-apple, hid-gamepad, hid-ite,
> >   hid-keyboard, and hid-mouse passed.
> > - The full HID suite was not completed; it stopped during hid-
> > multitouch.
> > - The #syz test request was rejected because syzbot's automated
> > patch
> >   testing is temporarily unavailable.
> 
> None of the above tested devices are HID SENSOR HUB related, so need
> some tests.
> Hi Lixu,
> 
> Please give a try.
> 
> Thanks,
> Srinivas
> 
> > 
> > Fixes: 8a5b38c3fd70 ("HID: hid-sensor-hub: don't use stale
> > platform-
> > data on remove")
> > Reported-by: syzbot+fb951970a334aaff00ba@syzkaller.appspotmail.com
> > Closes:
> > https://syzkaller.appspot.com/bug?extid=fb951970a334aaff00ba
> > Signed-off-by: Zheng Shuhao <viennaoffh@gmail.com>
> > ---
> > Changes in v2:
> > - Pass sensor_data to the callback instead of NULL.
> > - Skip children without Sensor Hub platform_data; add a battery-
> > child
> > test.
> > - Add Fixes and correct the syzbot address; remove the self-added
> > Tested-by.
> > 
> >  drivers/hid/hid-sensor-hub.c | 12 ++++++++----
> >  1 file changed, 8 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/hid/hid-sensor-hub.c b/drivers/hid/hid-sensor-
> > hub.c
> > index 6470a290ebfc5..4c447f62480b5 100644
> > --- a/drivers/hid/hid-sensor-hub.c
> > +++ b/drivers/hid/hid-sensor-hub.c
> > @@ -804,9 +804,16 @@ static int sensor_hub_probe(struct hid_device
> > *hdev,
> >  static int sensor_hub_finalize_pending_fn(struct device *dev, void
> > *data)
> >  {
> >  	struct hid_sensor_hub_device *hsdev = dev->platform_data;
> > +	struct sensor_hub_data *sensor_data = data;
> > +	unsigned long flags;
> > +
> > +	if (!hsdev)
> > +		return 0;
> >  
> > +	spin_lock_irqsave(&sensor_data->lock, flags);
> >  	if (hsdev->pending.status)
> >  		complete(&hsdev->pending.ready);
> > +	spin_unlock_irqrestore(&sensor_data->lock, flags);
> >  
> >  	return 0;
> >  }
> > @@ -814,16 +821,13 @@ static int
> > sensor_hub_finalize_pending_fn(struct device *dev, void *data)
> >  static void sensor_hub_remove(struct hid_device *hdev)
> >  {
> >  	struct sensor_hub_data *data = hid_get_drvdata(hdev);
> > -	unsigned long flags;
> >  
> >  	hid_dbg(hdev, " hardware removed\n");
> >  	hid_hw_close(hdev);
> >  	hid_hw_stop(hdev);
> >  
> > -	spin_lock_irqsave(&data->lock, flags);
> > -	device_for_each_child(&hdev->dev, NULL,
> > +	device_for_each_child(&hdev->dev, data,
> >  			      sensor_hub_finalize_pending_fn);
> > -	spin_unlock_irqrestore(&data->lock, flags);
> >  
> >  	mfd_remove_devices(&hdev->dev);
> >  	mutex_destroy(&data->mutex);

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH v2] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 15:30     ` Srinivas Pandruvada
@ 2026-11-20 17:05       ` Zheng Shuhao
  2026-11-20 17:56         ` Srinivas Pandruvada
  0 siblings, 1 reply; 13+ messages in thread
From: Zheng Shuhao @ 2026-11-20 17:05 UTC (permalink / raw)
  To: Srinivas Pandruvada
  Cc: Jiri Kosina, Jonathan Cameron, Benjamin Tissoires, Heiko Stuebner,
	Lee Jones, linux-iio, linux-kernel, Ayaan Mirza Baig,
	syzbot+fb951970a334aaff00ba, Zhang Lixu

 > In reality this can't happen as the hid_hw_close() and hid_hw_stop()
> is called before, so there can't be any raw_event.

It might not be reachable after hid_hw_stop(). I may have misunderstood the
lockdep report and overstated this as a possible deadlock. Sorry.

> How come this is possible? The HID sensor hub is only deleting its own
> children. Can you give an example here?

With CONFIG_HID_BATTERY_STRENGTH=y, this UHID device creates a Sensor Hub and
a Joystick battery. The HID input layer registers the battery power supply as
a child of the HID device, without platform_data.

With the unpatched hid-sensor-hub module, UHID_DESTROY triggers a KASAN
NULL-pointer dereference in sensor_hub_finalize_pending_fn():

  KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
  RIP: sensor_hub_finalize_pending_fn+0x44/0x65 [hid_sensor_hub]
  Call trace: device_for_each_child -> sensor_hub_remove -> hid_device_remove
              -> hid_destroy_device -> uhid_dev_destroy -> uhid_char_write

The reproducer is below:

#define _GNU_SOURCE
#include <dirent.h>
#include <errno.h>
#include <fcntl.h>
#include <linux/input.h>
#include <linux/uhid.h>
#include <poll.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>

static const unsigned char sensor_battery_rdesc[] = {
	0x05, 0x20,       /* Usage Page (Sensors) */
	0x09, 0x10,       /* Usage (Sensor Hub) */
	0xa1, 0x01,       /* Collection (Application) */
	0xc0,             /* End Collection */
	0x05, 0x01,       /* Usage Page (Generic Desktop) */
	0x09, 0x04,       /* Usage (Joystick) */
	0xa1, 0x01,       /* Collection (Application) */
	0x05, 0x06,       /* Usage Page (Generic Device Controls) */
	0x09, 0x20,       /* Usage (Battery Strength) */
	0x15, 0x00,       /* Logical Minimum (0) */
	0x25, 0x64,       /* Logical Maximum (100) */
	0x75, 0x08,       /* Report Size (8) */
	0x95, 0x01,       /* Report Count (1) */
	0x81, 0x02,       /* Input (Data, Variable, Absolute) */
	0xc0,             /* End Collection */
};

static int send_event(int fd, struct uhid_event *event)
{
	ssize_t written = write(fd, event, sizeof(*event));

	if (written == sizeof(*event))
		return 0;
	if (written < 0)
		perror("write /dev/uhid");
	else
		fprintf(stderr, "short UHID write: %zd\n", written);
	return -1;
}

static bool hid_battery_exists(void)
{
	static const char prefix[] = "hid-0003:1234:5678.";
	DIR *dir = opendir("/sys/class/power_supply");
	struct dirent *entry;
	bool found = false;

	if (!dir)
		return false;
	while ((entry = readdir(dir))) {
		if (!strncmp(entry->d_name, prefix, sizeof(prefix) - 1)) {
			found = true;
			break;
		}
	}
	closedir(dir);
	return found;
}

int main(void)
{
	struct uhid_event event = { 0 };
	struct pollfd pfd;
	int fd, attempt;
	bool started = false;
	bool battery_found = false;

	fd = open("/dev/uhid", O_RDWR | O_CLOEXEC);
	if (fd < 0) {
		perror("open /dev/uhid");
		return 1;
	}

	event.type = UHID_CREATE2;
	strcpy((char *)event.u.create2.name, "sensor-hub-battery-null-test");
	strcpy((char *)event.u.create2.phys, "hid-null-test");
	event.u.create2.rd_size = sizeof(sensor_battery_rdesc);
	event.u.create2.bus = BUS_USB;
	event.u.create2.vendor = 0x1234;
	event.u.create2.product = 0x5678;
	memcpy(event.u.create2.rd_data, sensor_battery_rdesc,
	       sizeof(sensor_battery_rdesc));
	if (send_event(fd, &event))
		goto out_close;

	pfd.fd = fd;
	pfd.events = POLLIN;
	if (poll(&pfd, 1, 5000) <= 0) {
		fprintf(stderr, "timed out waiting for UHID_START\n");
		goto out_destroy;
	}
	memset(&event, 0, sizeof(event));
	if (read(fd, &event, sizeof(event)) != sizeof(event)) {
		perror("read UHID event");
		goto out_destroy;
	}
	started = event.type == UHID_START;
	if (!started) {
		fprintf(stderr, "unexpected UHID event: %u\n", event.type);
		goto out_destroy;
	}

	for (attempt = 0; attempt < 50 && !hid_battery_exists(); attempt++)
		usleep(100000);
	if (!hid_battery_exists()) {
		fprintf(stderr, "HID battery power-supply child was not created\n");
		goto out_destroy;
	}
	battery_found = true;
	puts("HID battery power-supply child created");

out_destroy:
	memset(&event, 0, sizeof(event));
	event.type = UHID_DESTROY;
	if (send_event(fd, &event))
		goto out_close;
	usleep(300000);
	if (!started)
		goto out_close;
	puts("UHID device destroyed without userspace error");
	close(fd);
	return battery_found ? 0 : 1;

out_close:
	close(fd);
	return 1;
}

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [PATCH v2] HID: sensor-hub: Avoid holding spinlock across child iteration
  2026-11-20 17:05       ` Zheng Shuhao
@ 2026-11-20 17:56         ` Srinivas Pandruvada
  0 siblings, 0 replies; 13+ messages in thread
From: Srinivas Pandruvada @ 2026-11-20 17:56 UTC (permalink / raw)
  To: Zheng Shuhao
  Cc: Jiri Kosina, Jonathan Cameron, Benjamin Tissoires, Heiko Stuebner,
	Lee Jones, linux-iio, linux-kernel, Ayaan Mirza Baig,
	syzbot+fb951970a334aaff00ba, Zhang Lixu

On Sat, 2026-11-21 at 01:05 +0800, Zheng Shuhao wrote:
>  > In reality this can't happen as the hid_hw_close() and
> hid_hw_stop()
> > is called before, so there can't be any raw_event.
> 
> It might not be reachable after hid_hw_stop(). I may have
> misunderstood the
> lockdep report and overstated this as a possible deadlock. Sorry.
> 
> > How come this is possible? The HID sensor hub is only deleting its
> > own
> > children. Can you give an example here?
> 
> With CONFIG_HID_BATTERY_STRENGTH=y, this UHID device creates a Sensor
> Hub and
> a Joystick battery. The HID input layer registers the battery power
> supply as
> a child of the HID device, without platform_data.
> 
> With the unpatched hid-sensor-hub module, UHID_DESTROY triggers a
> KASAN
> NULL-pointer dereference in sensor_hub_finalize_pending_fn():
> 
>   KASAN: null-ptr-deref in range [0x0000000000000028-
> 0x000000000000002f]
>   RIP: sensor_hub_finalize_pending_fn+0x44/0x65 [hid_sensor_hub]
>   Call trace: device_for_each_child -> sensor_hub_remove ->
> hid_device_remove
>               -> hid_destroy_device -> uhid_dev_destroy ->
> uhid_char_write
> 

OK, Adding to commit description will help for others.

Thanks,
Srinivas

> The reproducer is below:
> 
> #define _GNU_SOURCE
> #include <dirent.h>
> #include <errno.h>
> #include <fcntl.h>
> #include <linux/input.h>
> #include <linux/uhid.h>
> #include <poll.h>
> #include <stdbool.h>
> #include <stdio.h>
> #include <string.h>
> #include <unistd.h>
> 
> static const unsigned char sensor_battery_rdesc[] = {
> 	0x05, 0x20,       /* Usage Page (Sensors) */
> 	0x09, 0x10,       /* Usage (Sensor Hub) */
> 	0xa1, 0x01,       /* Collection (Application) */
> 	0xc0,             /* End Collection */
> 	0x05, 0x01,       /* Usage Page (Generic Desktop) */
> 	0x09, 0x04,       /* Usage (Joystick) */
> 	0xa1, 0x01,       /* Collection (Application) */
> 	0x05, 0x06,       /* Usage Page (Generic Device Controls) */
> 	0x09, 0x20,       /* Usage (Battery Strength) */
> 	0x15, 0x00,       /* Logical Minimum (0) */
> 	0x25, 0x64,       /* Logical Maximum (100) */
> 	0x75, 0x08,       /* Report Size (8) */
> 	0x95, 0x01,       /* Report Count (1) */
> 	0x81, 0x02,       /* Input (Data, Variable, Absolute) */
> 	0xc0,             /* End Collection */
> };
> 
> static int send_event(int fd, struct uhid_event *event)
> {
> 	ssize_t written = write(fd, event, sizeof(*event));
> 
> 	if (written == sizeof(*event))
> 		return 0;
> 	if (written < 0)
> 		perror("write /dev/uhid");
> 	else
> 		fprintf(stderr, "short UHID write: %zd\n", written);
> 	return -1;
> }
> 
> static bool hid_battery_exists(void)
> {
> 	static const char prefix[] = "hid-0003:1234:5678.";
> 	DIR *dir = opendir("/sys/class/power_supply");
> 	struct dirent *entry;
> 	bool found = false;
> 
> 	if (!dir)
> 		return false;
> 	while ((entry = readdir(dir))) {
> 		if (!strncmp(entry->d_name, prefix, sizeof(prefix) -
> 1)) {
> 			found = true;
> 			break;
> 		}
> 	}
> 	closedir(dir);
> 	return found;
> }
> 
> int main(void)
> {
> 	struct uhid_event event = { 0 };
> 	struct pollfd pfd;
> 	int fd, attempt;
> 	bool started = false;
> 	bool battery_found = false;
> 
> 	fd = open("/dev/uhid", O_RDWR | O_CLOEXEC);
> 	if (fd < 0) {
> 		perror("open /dev/uhid");
> 		return 1;
> 	}
> 
> 	event.type = UHID_CREATE2;
> 	strcpy((char *)event.u.create2.name, "sensor-hub-battery-
> null-test");
> 	strcpy((char *)event.u.create2.phys, "hid-null-test");
> 	event.u.create2.rd_size = sizeof(sensor_battery_rdesc);
> 	event.u.create2.bus = BUS_USB;
> 	event.u.create2.vendor = 0x1234;
> 	event.u.create2.product = 0x5678;
> 	memcpy(event.u.create2.rd_data, sensor_battery_rdesc,
> 	       sizeof(sensor_battery_rdesc));
> 	if (send_event(fd, &event))
> 		goto out_close;
> 
> 	pfd.fd = fd;
> 	pfd.events = POLLIN;
> 	if (poll(&pfd, 1, 5000) <= 0) {
> 		fprintf(stderr, "timed out waiting for
> UHID_START\n");
> 		goto out_destroy;
> 	}
> 	memset(&event, 0, sizeof(event));
> 	if (read(fd, &event, sizeof(event)) != sizeof(event)) {
> 		perror("read UHID event");
> 		goto out_destroy;
> 	}
> 	started = event.type == UHID_START;
> 	if (!started) {
> 		fprintf(stderr, "unexpected UHID event: %u\n",
> event.type);
> 		goto out_destroy;
> 	}
> 
> 	for (attempt = 0; attempt < 50 && !hid_battery_exists();
> attempt++)
> 		usleep(100000);
> 	if (!hid_battery_exists()) {
> 		fprintf(stderr, "HID battery power-supply child was
> not created\n");
> 		goto out_destroy;
> 	}
> 	battery_found = true;
> 	puts("HID battery power-supply child created");
> 
> out_destroy:
> 	memset(&event, 0, sizeof(event));
> 	event.type = UHID_DESTROY;
> 	if (send_event(fd, &event))
> 		goto out_close;
> 	usleep(300000);
> 	if (!started)
> 		goto out_close;
> 	puts("UHID device destroyed without userspace error");
> 	close(fd);
> 	return battery_found ? 0 : 1;
> 
> out_close:
> 	close(fd);
> 	return 1;
> }

^ permalink raw reply	[flat|nested] 13+ messages in thread

end of thread, other threads:[~2026-11-20 17:56 UTC | newest]

Thread overview: 13+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-20  7:56 [PATCH] HID: sensor-hub: Avoid holding spinlock across child iteration Zheng Shuhao
2026-11-20  9:34 ` sashiko-bot
2026-11-20 10:50 ` Joshua Crofts
2026-11-20 10:50   ` syzbot
2026-11-20 11:09   ` Joshua Crofts
2026-11-20 11:09     ` syzbot
2026-11-20 12:14 ` Srinivas Pandruvada
2026-11-20 13:49   ` Zheng Shuhao
2026-11-20 13:50 ` [PATCH v2] " Zheng Shuhao
2026-11-20 15:12   ` Srinivas Pandruvada
2026-11-20 15:30     ` Srinivas Pandruvada
2026-11-20 17:05       ` Zheng Shuhao
2026-11-20 17:56         ` Srinivas Pandruvada

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.