kernel-packages team mailing list archive
-
kernel-packages team
-
Mailing list archive
-
Message #101754
[Bug 1413992] Re: Kernel oopses on access to address 0x8 when cdc-acm device is inserted with invalid descriptor.
** Changed in: linux (Ubuntu)
Status: Triaged => In Progress
** Changed in: hwe-next/vivid
Status: Triaged => In Progress
** Changed in: hwe-next/utopic
Status: New => In Progress
** Changed in: hwe-next/trusty
Status: New => In Progress
** Changed in: hwe-next/trusty
Importance: Undecided => High
** Changed in: hwe-next/utopic
Importance: Undecided => High
** Changed in: hwe-next/trusty
Assignee: (unassigned) => Adam Lee (adam8157)
** Changed in: hwe-next/utopic
Assignee: (unassigned) => Adam Lee (adam8157)
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1413992
Title:
Kernel oopses on access to address 0x8 when cdc-acm device is inserted
with invalid descriptor.
Status in HWE Next Project:
In Progress
Status in HWE Next trusty series:
In Progress
Status in HWE Next utopic series:
In Progress
Status in HWE Next vivid series:
In Progress
Status in linux package in Ubuntu:
In Progress
Bug description:
Invalid configuration descriptor as follows:
#+BEGIN_SRC text
0000 09 02 43 00 02 01 00 80 64 09 04 00 00 01 02 02 ..C.....d.......
0010 00 00 05 24 00 10 01 04 24 02 06 04 24 01 00 01 ...$....$...$...
0020 05 24 06 00 01 07 05 81 03 08 00 ff 09 04 01 00 .$..............
0030 02 0a 00 00 00 07 05 82 02 40 00 ff 07 05 01 02 .........@......
0040 20 00 ff ..
#+END_SRC text
In particular, the CDC Call Management Descriptor has its length
declared too short (4 instead of 5), and the following CDC Union
Descriptor is therefore unreachable.
*** Code problems:
1. The ~while (buflen > 0)~ loop that parses the interface aux data
does not perform correct boundary checking. In the above case,
~call_interface_num = buffer[4];~ accesses outside of the
(declared) descriptor content.
2. If a union header is missing, there is no code path that checks
whether the ~data_interface~ (resolved from ~call_interface_num~)
actually exists. Later ~if
(data_interface->cur_altsetting->desc.bInterfaceClass~ dereferences
~data_interface~.
ref: https://bugzilla.kernel.org/show_bug.cgi?id=83551
issue 2 was already fixed, issue 1's fix is in progress of upstream
merging, open this bug to track.
To manage notifications about this bug go to:
https://bugs.launchpad.net/hwe-next/+bug/1413992/+subscriptions
References