Re: weird use-after-free bug in module_put

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

Hi Dave,

On Fri, Oct 19, 2012 at 11:34:52AM -0400, Dave Jones wrote:
> On Fri, Oct 19, 2012 at 10:43:51AM -0400, Dave Jones wrote:
>  > I've hit this twice in the last two days while fuzz testing.
>  > (Both times on i686 only, my x86-64 tests aren't hitting it
>  >  for some reason).
>  > 
>  > BUG: unable to handle kernel paging request at 6b6b6ce3
>  > IP: [<c10b52fe>] module_put+0x1e/0x160
>  > *pdpt = 0000000025a4b001 *pde = 0000000000000000 
>  > Oops: 0000 [#1] PREEMPT SMP 
>  > Modules linked in: fuse tun binfmt_misc nfnetlink nfc caif_socket caif phonet bluetooth rfkill can llc2 pppoe pppox ppp_generic slhc irda crc_ccitt rds af_key decnet rose x25 atm netrom appletalk ipx p8023 psnap p8022 llc ax25 lockd sunrpc ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 nf_conntrack_ipv4 ip6table_filter ip6_tables nf_defrag_ipv4 xt_state nf_conntrack kvm_intel kvm microcode serio_raw pcspkr i2c_i801 tg3 i2c_core shpchp raid0 ata_piix
>  > Pid: 512, comm: acpid Not tainted 3.7.0-rc1+ #11 Dell Inc.                 Precision WorkStation 490    /0DT031
>  > EIP: 0060:[<c10b52fe>] EFLAGS: 00010246 CPU: 1
>  > EIP is at module_put+0x1e/0x160
>  > EAX: 00000000 EBX: 6b6b6b6b ECX: 00000000 EDX: c118509c
>  > ESI: 00000010 EDI: 6b6b6b6b EBP: e5ae9f44 ESP: e5ae9f34
>  >  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
>  > CR0: 8005003b CR2: 6b6b6ce3 CR3: 25a4a000 CR4: 000007f0
>  > DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
>  > DR6: ffff0ff0 DR7: 00000400
>  > Process acpid (pid: 512, ti=e5ae8000 task=e6311680 task.ti=e5ae8000)
>  > Stack:
>  >  e6062140 6b6b6b6b 00000010 f01ce540 e5ae9f50 c118509c e6062140 e5ae9f80
>  >  c11821ed 00000001 00000000 00000000 f2073410 ef256814 ef256814 e6062148
>  >  00000000 e6311a60 e6311680 e5ae9f88 c118226d e5ae9f9c c1062f19 00000002
>  > Call Trace:
>  >  [<c118509c>] cdev_put+0x1c/0x20
>  >  [<c11821ed>] __fput+0x20d/0x280
>  >  [<c118226d>] ____fput+0xd/0x10
>  >  [<c1062f19>] task_work_run+0x89/0xb0
>  >  [<c1002c41>] do_notify_resume+0x61/0xa0
>  >  [<c15d32f0>] work_notifysig+0x29/0x31
>  > Code: 51 00 eb df 89 f6 8d bc 27 00 00 00 00 55 89 e5 57 56 53 83 ec 04 66 66 66 66 90 85 c0 89 c7 74 44 b8 01 00 00 00 e8 c2 14 52 00 <8b> 87 78 01 00 00 64 ff 40 04 8b 45 04 89 45 f0 66 66 66 66 90
>  > 
>  > 
>  > It looks like the chardev went away under our feet.
>  > How can this happen ?

We are now removing instance of character device corresponding to input
device when input device disappears.

Ah, I know... cdev is embedded in evdev, but lives longer.. I do want to
keep cdev embedded as it allows me to easily get to evdev in
evdev_open(), but I need to be able to add and then drop reference to
evdev from cdev's ->release() method. This means I need to override it.

Or I could have cdev separately allocated, but then I'd like to have a
void pointer in "struct cdev" so I could get from it back to
corresponding evdev.

Al, any recommendation?

Thanks!

-- 
Dmitry