On Fri, 31 Oct 2008, Helge Deller wrote:

After sucessful bootup (without any USB devices attached)
I get this when I insert a USB keyboard:
---------------
usb 1-1: new low speed USB device using ohci_hcd and address 2
usb 1-1: configuration #1 chosen from 1 choice
input: SILITEK USB Keyboard and Mouse as /class/input/input0
Slab corruption: size-4096 start=8dd9b000, len=4096
000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
040: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
050: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
generic-usb 0004:047B:0002.0001: input,hidraw0: USB HID v1.00 Keyboard [SILITEK USB Keyboard and Mouse] on usb-0000:00:0e.2-1/input0

Looks like something goes wrong somewhere around hid_connect() -- 
hidinput_connect() is called, then slab complains about corruption, and 
then the rest of the code in hid_connect() is executed, priting the 
'generic-usb ...' message.

I can't reproduce it here myself with CONFIG_DEBUG_SLAB=y. Could you 
please send me your config?

It is also quite strange that stacktrace is missing here for some 
reason.

On 2.6.28-rc1 I saw e.g. this:
--------------------
usbcore: registered new interface driver usbhid
usbhid: v2.6:USB HID core driver
usb 1-1: new low speed USB device using ohci_hcd and address 2
usb 1-1: configuration #1 chosen from 1 choice
input: Logitech N48 as /class/input/input0
generic-usb 0003:046D:C001.0001: input,hidraw0: USB HID v1.00 Mouse 
[Logitech N48] on usb-0000:00:0e.2-1/inpu0
usb 1-1: New USB device found, idVendor=046d, idProduct=c001
usb 1-1: New USB device strings: Mfr=1, Product=2, SerialNumber=0
usb 1-1: Product: N48
usb 1-1: Manufacturer: Logitech
usb 1-2: new low speed USB device using ohci_hcd and address 3
usb 1-2: configuration #1 chosen from 1 choice
slab error in cache_alloc_debugcheck_after(): cache `size-512': double free, or memory outside object was oven
Backtrace:
   [<101a5724>] cache_alloc_debugcheck_after+0xd8/0x200
   [<101a5cac>] kmem_cache_alloc+0x1a0/0x1e8
   [<1042e294>] hid_register_report+0x60/0xc4
   [<1042e5f8>] hid_add_field+0x40/0x1a4
   [<1042ec40>] hid_parser_main+0x94/0xc4

This looks rather incomplete, backtrace starting at hid_parser_main is 
odd.

Also, this seems to happen for different slab cache than the previous 
corrutpion.

-- 
Jiri Kosina
SUSE Labs