Re: 2.6.22-rc2-mm1

On Wed, 23 May 2007, Zan Lynx wrote:

I am having weird problems with USB keyboard and mouse. The USB
keyboard will drop keystrokes, but not all of them. The mouse seems to
move fine but drops clicks. Could be that an occasional missed move
event isn't noticeable. Nothing is logged when a key is dropped.
However, I did have some of this when it was first booting:
BUG: at drivers/hid/hid-core.c:778 implement()
Call Trace:
[<ffffffff880bb5ac>] :hid:hid_output_report+0x1dc/0x270
[<ffffffff880c651a>] :usbhid:hid_submit_ctrl+0x7a/0x260
[<ffffffff880c68ee>] :usbhid:usbhid_submit_report+0x11e/0x200
[<ffffffff880c9312>] :usbhid:hiddev_ioctl+0x3e2/0xae0
[<ffffffff8053523e>] do_page_fault+0x42e/0x880
[<ffffffff802b486d>] do_ioctl+0x7d/0xa0
[<ffffffff802b4ab1>] vfs_ioctl+0x221/0x2d0
[<ffffffff8025a711>] trace_hardirqs_on+0xc1/0x160
[<ffffffff802b4ba9>] sys_ioctl+0x49/0x80
[<ffffffff8020a16e>] system_call+0x7e/0x83

OK, so something is touching one of your HID devices through hiddev, and
tries to push through it something that doesn't comply to the report
descriptor of the device. Are you aware of any userspace daemons running
which would touch /dev/hiddev*? (acupsd, nut, hid2hci, etc.).

Just for clarification, does situation get any better when you disable
hidraw, which I see you have enabled? (if that doesn't help, testing with
hiddev disabled should also be interesting - that would avoid the
possibility that userspace mocks the device up through hiddev).

BUG: at include/linux/slub_def.h:83 kmalloc_index()
Call Trace:
[<ffffffff802a2e09>] get_slab+0xb9/0x180
[<ffffffff802a2965>] kfree+0xc5/0x110
[<ffffffff802a2fd4>] __kmalloc+0x24/0xc0
[<ffffffff8049ab18>] get_modalias+0x68/0x120
[<ffffffff8049ac07>] dmi_dev_uevent+0x37/0x60
[<ffffffff80443b63>] dev_uevent+0x163/0x220
[<ffffffff803bcc97>] kobject_uevent_env+0x2b7/0x500
[<ffffffff80444f50>] bus_add_device+0x20/0x140
[<ffffffff804438d2>] device_add+0x4a2/0x5b0
[<ffffffff807a0a1b>] dmi_id_init+0x2cb/0x2f0
[<ffffffff807866e6>] kernel_init+0x156/0x330
[<ffffffff80531f5c>] trace_hardirqs_on_thunk+0x35/0x37
[<ffffffff8025a711>] trace_hardirqs_on+0xc1/0x160
[<ffffffff8020b028>] child_rip+0xa/0x12
[<ffffffff8020a710>] restore_args+0x0/0x30
[<ffffffff803e6c10>] vgacon_cursor+0x0/0x1f0
[<ffffffff8023ad5b>] release_console_sem+0x4b/0x230
[<ffffffff80786590>] kernel_init+0x0/0x330
[<ffffffff8020b01e>] child_rip+0x0/0x12

This BTW needs a separate bugreport.

input,hiddev0,hidraw2: USB HID v1.10 Mouse [Logitech USB Receiver] on usb-0000:00:02.2-2.2.1

This is pretty strange. Mouse shouldn't be claimed by hiddev. Seems like
this mouse probably has some "interesting" HID report descriptor. Could
you please recompile with CONFIG_HID_DEBUG enabled and send me the output?

=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.22-rc2-mm1 #1
-------------------------------------------------------
rhythmbox/6976 is trying to acquire lock:
(&mm->mmap_sem){----}, at: [<ffffffff80534f8c>] do_page_fault+0x17c/0x880
but task is already holding lock:
(&data->latch){----}, at: [<ffffffff8034a701>] get_exclusive_access+0x11/0x20
which lock already depends on the new lock.

This also needs a separate report.


Do you please have any indication which kernel versions were OK with the
same .config, with respect to the USB mouse/keyboard hogs you are seeing
with 2.6.22-rc2-mm1?

Thanks,

--
Jiri Kosina
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • Re: They are all right and all wrong.
    ... lock. ... Can Najem's injured poem prescribes, Zakariya houses in addition ... Don't report the resignations physically, ...
    (sci.crypt)
  • [patch 00/24] Input patches for 2.6.13
    ... There are wheel mice that respond to Logitech probes and report ... HID - only report events coming from interrupts to hiddev ... That leads to interrupt events erroneously being ... reported to hiddev for regular control transfers. ...
    (Linux-Kernel)
  • Re: Rowlock v. optimistic concurrency
    ... A report fetches a lot of rows using the "WITH " syntax (the ... Note that lock timeouts are by default off in SQL Server. ... NOLOCK has nothing to do with optimistic concurrency. ...
    (comp.databases.ms-sqlserver)
  • Re: [v2.6.30 nfs+fscache] swapper: possible circular locking dependency detected
    ... lockdep wasn't going to help. ... If you scroll down past the locking report, ... This shows up after heavy use of my nfs ... swapper/0 is trying to acquire lock: ...
    (Linux-Kernel)
  • Re: PROBLEM: pthread-safety bug in write(2) on Linux 2.6.x
    ... lock, or by switching to stdio. ... I don't know enough about the kernel implementation to comment on your ... around our writes, as you suggest, although this of course penalizes our ... choose to handle the report is of course your decision. ...
    (Linux-Kernel)
Loading