linux/drivers/input/misc/uinput.c, branch v4.14

Input: uinput - avoid crash when sending FF request to device going away

2017-09-21T23:31:23Z

If FF request comes in while uinput device is going away, uinput_request_send() will fail with -ENODEV, and uinput_request_submit() will attempt to mark the slot as unused by calling uinput_request_done(). Unfortunately in this case we haven't initialized request->done completion yet, and we get a crash: [ 39.402036] BUG: spinlock bad magic on CPU#1, fftest/3108 [ 39.402046] lock: 0xffff88006a93bb00, .magic: 00000000, .owner: /39, .owner_cpu: 1217155072 [ 39.402055] CPU: 1 PID: 3108 Comm: fftest Tainted: G W 4.13.0+ #15 [ 39.402059] Hardware name: LENOVO 20HQS0EG02/20HQS0EG02, BIOS N1MET37W (1.22 ) 07/04/2017 [ 39.402064] 0000000000000086 f0fad82f3ceaa120 ffff88006a93b9a0 ffffffff9de941bb [ 39.402077] ffff88026df8ae00 ffff88006a93bb00 ffff88006a93b9c0 ffffffff9dca62b7 [ 39.402088] ffff88006a93bb00 ffff88006a93baf8 ffff88006a93b9e0 ffffffff9dca62e7 [ 39.402099] Call Trace: [ 39.402112] [] dump_stack+0x4d/0x63 [ 39.402123] [] spin_dump+0x97/0x9c [ 39.402130] [] spin_bug+0x2b/0x2d [ 39.402138] [] do_raw_spin_lock+0x28/0xfd [ 39.402147] [] _raw_spin_lock_irqsave+0x19/0x1f [ 39.402154] [] complete+0x1d/0x48 [ 39.402162] [] 0xffffffffc04f30af [ 39.402167] [] 0xffffffffc04f468c [ 39.402177] [] ? __slab_free+0x22f/0x359 [ 39.402184] [] ? tk_clock_read+0xc/0xe [ 39.402189] [] 0xffffffffc04f471f [ 39.402195] [] ? __wake_up+0x44/0x4b [ 39.402200] [] ? 0xffffffffc04f3240 [ 39.402207] [] erase_effect+0xa1/0xd2 [ 39.402214] [] input_ff_flush+0x43/0x5c [ 39.402219] [] 0xffffffffc04f32ad [ 39.402227] [] input_flush_device+0x3d/0x51 [ 39.402234] [] evdev_flush+0x49/0x5c [ 39.402243] [] filp_close+0x3f/0x65 [ 39.402253] [] put_files_struct+0x66/0xc1 [ 39.402261] [] exit_files+0x47/0x4e [ 39.402270] [] do_exit+0x483/0x969 [ 39.402278] [] ? recalc_sigpending_tsk+0x3d/0x44 [ 39.402285] [] do_group_exit+0x42/0xb0 [ 39.402293] [] get_signal+0x58d/0x5bf [ 39.402300] [] do_signal+0x37/0x53e [ 39.402307] [] ? evdev_ioctl_handler+0xac8/0xb04 [ 39.402314] [] ? evdev_ioctl+0x10/0x12 [ 39.402321] [] ? do_vfs_ioctl+0x42e/0x501 [ 39.402328] [] prepare_exit_to_usermode+0x66/0x90 [ 39.402333] [] syscall_return_slowpath+0xe3/0xec [ 39.402339] [] int_ret_from_sys_call+0x25/0x8f While we could solve this by simply initializing the completion earlier, we are better off rearranging the code a bit so we avoid calling complete() on requests that we did not send out. This patch consolidates marking request slots as free in one place (in uinput_request_submit(), the same place where we acquire them) and having everyone else simply signal completion of the requests. Fixes: 00ce756ce53a ("Input: uinput - mark failed submission requests as free") Signed-off-by: Dmitry Torokhov

Input: uinput - avoid FF flush when destroying device

2017-09-21T23:31:22Z

Normally, when input device supporting force feedback effects is being destroyed, we try to "flush" currently playing effects, so that the physical device does not continue vibrating (or executing other effects). Unfortunately this does not work well for uinput as flushing of the effects deadlocks with the destroy action: - if device is being destroyed because the file descriptor is being closed, then there is noone to even service FF requests; - if device is being destroyed because userspace sent UI_DEV_DESTROY, while theoretically it could be possible to service FF requests, userspace is unlikely to do so (they'd need to make sure FF handling happens on a separate thread) even if kernel solves the issue with FF ioctls deadlocking with UI_DEV_DESTROY ioctl on udev->mutex. To avoid lockups like the one below, let's install a custom input device flush handler, and avoid trying to flush force feedback effects when we destroying the device, and instead rely on uinput to shut off the device properly. NMI watchdog: Watchdog detected hard LOCKUP on cpu 3 ... <> [] _raw_spin_lock_irqsave+0x37/0x40 [] complete+0x1d/0x50 [] uinput_request_done+0x3c/0x40 [uinput] [] uinput_request_submit.part.7+0x47/0xb0 [uinput] [] uinput_dev_erase_effect+0x5b/0x76 [uinput] [] erase_effect+0xad/0xf0 [] flush_effects+0x4d/0x90 [] input_flush_device+0x40/0x60 [] evdev_cleanup+0xac/0xc0 [] evdev_disconnect+0x2b/0x60 [] __input_unregister_device+0xac/0x150 [] input_unregister_device+0x47/0x70 [] uinput_destroy_device+0xb5/0xc0 [uinput] [] uinput_ioctl_handler.isra.9+0x65e/0x740 [uinput] [] ? do_futex+0x12b/0xad0 [] uinput_ioctl+0x18/0x20 [uinput] [] do_vfs_ioctl+0x298/0x480 [] ? security_file_ioctl+0x43/0x60 [] SyS_ioctl+0x79/0x90 [] entry_SYSCALL_64_fastpath+0x12/0x71 Reported-by: Rodrigo Rivas Costa Reported-by: Clément VUCHENER Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=193741 Signed-off-by: Dmitry Torokhov

Input: uinput - fix crash when mixing old and new init style

2017-02-01T18:33:11Z

If user tries to initialize uinput device mixing old and new style initialization (i.e. using old UI_SET_ABSBIT instead of UI_ABS_SETUP, we forget to allocate input->absinfo and will crash when trying to send absolute events: ioctl(ui, UI_DEV_SETUP, &us); ioctl(ui, UI_SET_PHYS, "Test"); ioctl(ui, UI_SET_EVBIT, EV_ABS); ioctl(ui, UI_SET_ABSBIT, ABS_X); ioctl(ui, UI_SET_ABSBIT, ABS_Y); ioctl(ui, UI_DEV_CREATE, 0); Reported-by: Rodrigo Rivas Costa Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=191811 Fixes: fbae10db0940 ("Input: uinput - rework ABS validation") Reviewed-by: Benjamin Tissoires Cc: stable@vger.kernel.org Signed-off-by: Dmitry Torokhov

miscdevice: Add helper macro for misc device boilerplate

2016-08-31T12:12:35Z

Many modules call misc_register and misc_deregister in its module init and exit methods without any additional code. This ends up being boilerplate. This patch adds helper macro module_misc_device(), that replaces module_init()/ module_exit() with template functions. This patch also converts drivers to use new macro. Change since v1: Add device.h include in miscdevice.h as module_driver macro was not available from other include files in some architectures. Signed-off-by: PrasannaKumar Muralidharan Signed-off-by: Greg Kroah-Hartman

Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input

2016-05-28T02:14:35Z

Pull more input subsystem updates from Dmitry Torokhov: "Just a few more driver fixes; new drivers will be coming in the next merge window" * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input: Input: pwm-beeper - fix - scheduling while atomic Input: xpad - xbox one elite controller support Input: xpad - add more third-party controllers Input: xpad - prevent spurious input from wired Xbox 360 controllers Input: xpad - move pending clear to the correct location Input: uinput - handle compat ioctl for UI_SET_PHYS

Input: uinput - handle compat ioctl for UI_SET_PHYS

2016-05-20T18:00:43Z

When running a 32-bit userspace on a 64-bit kernel, the UI_SET_PHYS ioctl needs to be treated with special care, as it has the pointer size encoded in the command. Signed-off-by: Ricky Liang Cc: stable@vger.kernel.org Signed-off-by: Dmitry Torokhov

drivers/input: eliminate INPUT_COMPAT_TEST macro

2016-03-25T23:37:42Z

INPUT_COMPAT_TEST became much simpler after commit f4056b52845283 ("input: redefine INPUT_COMPAT_TEST as in_compat_syscall()") so we can cleanly eliminate it altogether. Acked-by: Dmitry Torokhov Cc: Andy Lutomirski Cc: Andy Lutomirski Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Input: uinput - sanity check on ff_effects_max and EV_FF

2015-12-19T01:48:52Z

Currently the user can set ff_effects_max to zero with the EV_FF bit (and the FF_GAIN and/or FF_AUTOCENTER bits) set, in this case the uninitialized methods ff->set_gain and/or ff->set_autocenter can be dereferenced, resulting in a kernel oops. Check in uinput_create_device() and print a helpful message and return -EINVAL in case the check fails. Signed-off-by: Elias Vanderstuyft Signed-off-by: Dmitry Torokhov

Input: uinput - rework ABS validation

2015-12-19T01:48:51Z

Rework the uinput ABS validation to check passed absinfo data immediately, but do ABS initialization as last step in UI_DEV_CREATE. The behavior observed by user-space is not changed, as ABS initialization was never checked for errors. With this in place, the order of device initialization and abs configuration is no longer fixed. Userspace can initialize the device and afterwards set absinfo just fine. Signed-off-by: David Herrmann Reviewed-by: Benjamin Tissoires Tested-by: Benjamin Tissoires Signed-off-by: Dmitry Torokhov

Input: uinput - add new UINPUT_DEV_SETUP and UI_ABS_SETUP ioctl

2015-12-19T01:48:50Z

This adds two new ioctls, UINPUT_DEV_SETUP and UI_ABS_SETUP, that replaces the old device setup method (by write()'ing "struct uinput_user_dev" to the node). The old method is not easily extendable and requires huge payloads. Furthermore, overloading write() without properly versioned objects is error-prone. Therefore, we introduce two new ioctls to replace the old method. These ioctls support all features of the old method, plus a "resolution" field for absinfo. Furthermore, it's properly forward-compatible to new ABS codes and a growing "struct input_absinfo" structure. UI_ABS_SETUP also allows user-space to skip unknown axes if not set. There is no need to copy the whole array temporarily into the kernel, but instead the caller issues several ioctl where we copy each value manually. Signed-off-by: David Herrmann Signed-off-by: Benjamin Tissoires Reviewed-by: David Herrmann Signed-off-by: Dmitry Torokhov