Hi
On Sun, Nov 17, 2013 at 10:36 AM, Sven Eckelmann <sven@xxxxxxxxxxxxx> wrote:
> On Saturday 16 November 2013 17:30:25 simon@xxxxxxxxxxxxx wrote:
>> This patch appears to work OK with my DualShock/SixAxis controller (USB
>> connection), but causes a machine lockup when used with my Intec wired
>> controller.
>>
>> The Intec controller works OK up to the point when I start rumble,
>> sometimes the motor runs some times it doesn't. I am using SDL2's
>> 'testhaptic' application.
>
> Thanks a lot for this bug report. The testhaptic was a good testcase because
> it is a heavy user and can reproduce the problem quite easily. I've only
> tested it using testrumble and own programs which didn't seem to trigger the
> problem here. The problem is easy to explain:
>
> * usb_control_msg/usb_interrupt_msg/usb_bulk_msg/... may sleep
> * sony_play_effect may gets called in an softirq context (atomic)
>
> So it is a bad choice to use hid_output_raw_report (which calls
> usb_control_msg) in a ff_memless control function. I will just send a revert
> patch in some minutes.
Yeah, the input-ff callbacks cannot be handled inline. You also get
deadlocks with the input-spinlock. In the wiimote driver I simply
dispatch the ff-events to a workqueue. You can have a look at
drivers/hid/hid-wiimote-modules.c. You can get some ordering-problems
then, but these can usually be ignored as they just collapse events.
The related commit was:
commit f50f9aabf32db7414551ffdfdccc71be5f3d6e7d
Author: David Herrmann <dh.herrmann@xxxxxxxxx>
Date: Wed Oct 2 13:47:28 2013 +0200
HID: wiimote: fix FF deadlock
Thanks
David
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
