kevinmel2000 / google-security-research

Automatically exported from code.google.com/p/google-security-research
0 stars 0 forks source link

Windows Kernel ATMFD.DLL invalid memory access due to malformed CFF table (ATMFD+0x3440b / ATMFD+0x3440e) #384

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago
We have encountered a number of Windows kernel crashes in the ATMFD.DLL 
OpenType driver while processing corrupted OTF font files, such as:

---
PAGE_FAULT_IN_NONPAGED_AREA (50)
Invalid system memory was referenced.  This cannot be protected by try-except,
it must be protected by a Probe.  Typically the address is just plain bad or it
is pointing at freed memory.
Arguments:
Arg1: fff82008, memory referenced.
Arg2: 00000000, value 0 = read operation, 1 = write operation.
Arg3: 91a3440b, If non-zero, the instruction address which referenced the bad 
memory
    address.
Arg4: 00000000, (reserved)

Debugging Details:
------------------

*** ERROR: Module load completed but symbols could not be loaded for ATMFD.DLL

READ_ADDRESS:  fff82008 

FAULTING_IP: 
ATMFD+3440b
91a3440b 8b7e08          mov     edi,dword ptr [esi+8]

MM_INTERNAL_CODE:  0

IMAGE_NAME:  ATMFD.DLL

DEBUG_FLR_IMAGE_TIMESTAMP:  54e6a55a

MODULE_NAME: ATMFD

FAULTING_MODULE: 91a00000 ATMFD

DEFAULT_BUCKET_ID:  WIN7_DRIVER_FAULT

BUGCHECK_STR:  0x50

PROCESS_NAME:  csrss.exe

CURRENT_IRQL:  2

ANALYSIS_VERSION: 6.3.9600.17237 (debuggers(dbg).140716-0327) x86fre

TRAP_FRAME:  97ff8d54 -- (.trap 0xffffffff97ff8d54)
ErrCode = 00000000
eax=ff677018 ebx=00000001 ecx=00000001 edx=0000000b esi=fff82000 edi=fb63e940
eip=91a3440b esp=97ff8dc8 ebp=97ff8de8 iopl=0         nv up ei ng nz na pe nc
cs=0008  ss=0010  ds=0023  es=0023  fs=0030  gs=0000             efl=00010286
ATMFD+0x3440b:
91a3440b 8b7e08          mov     edi,dword ptr [esi+8] ds:0023:fff82008=????????
Resetting default scope

LAST_CONTROL_TRANSFER:  from 82724ce7 to 826c02d8

STACK_TEXT:  
97ff88a4 82724ce7 00000003 4b979438 00000065 nt!RtlpBreakWithStatusInstruction
97ff88f4 827257e5 00000003 00000000 00000000 nt!KiBugCheckDebugBreak+0x1c
97ff8cb8 826d3391 00000050 fff82008 00000000 nt!KeBugCheck2+0x68b
97ff8d3c 82685c48 00000000 fff82008 00000000 nt!MmAccessFault+0x104
97ff8d3c 91a3440b 00000000 fff82008 00000000 nt!KiTrap0E+0xdc
WARNING: Stack unwind information not available. Following frames may be wrong.
97ff8de8 91a345f7 fb6ba380 0000000b fb6ba3ac ATMFD+0x3440b
97ff8e10 91a29eec fb63e8c0 ff6770d8 91a463ec ATMFD+0x345f7
97ff8e34 91a2e987 fb8f4c70 91a463ec 00000f5c ATMFD+0x29eec
97ff9544 91a2f6e0 fb8f4c70 91a3f028 97ff9790 ATMFD+0x2e987
97ff9600 91a227ae fb8f4c70 91a3f028 97ff9790 ATMFD+0x2f6e0
97ff96ec 91a22858 fb8f4c70 97ff9790 97ff9814 ATMFD+0x227ae
97ff9718 91a132b2 fb8f4c70 91a3f028 97ff9790 ATMFD+0x22858
97ff987c 91a13689 ffffffff 97ff999c fb68af58 ATMFD+0x132b2
97ff98d0 91a0406d ffffffff 97ff999c 00000000 ATMFD+0x13689
97ff9924 91b2dcf2 ff7a5010 fb700cf0 00000001 ATMFD+0x406d
97ff996c 91b167cb ff7a5010 fb700cf0 00000001 win32k!PDEVOBJ::QueryFontData+0x3e
97ff99e0 91b41513 ffa6a130 fb93cb14 000000e0 win32k!xInsertMetricsRFONTOBJ+0x9c
97ff9a14 91b435f5 00000020 97ff9a3c 97ff9c74 
win32k!RFONTOBJ::bGetGlyphMetrics+0x131
97ff9cb8 91b56684 020101c3 00000100 00000020 win32k!GreGetCharABCWidthsW+0x147
97ff9d14 82682a66 020101c3 000000c0 00000040 win32k!NtGdiGetCharABCWidthsW+0xf8
97ff9d14 76ee70f4 020101c3 000000c0 00000040 nt!KiSystemServicePostCall
002cf224 00000000 00000000 00000000 00000000 ntdll!KiFastSystemCallRet

---

The crash represents a read from invalid memory; prior to being dereferenced as 
an address, the ESI register is loaded with a value from EAX+0x30 (on 32-bit 
Windows), which points into an "Adbe" pool allocation:

---
0: kd> !pool 9d619018
Pool page 9d619018 region is Paged session pool
9d619000 is not a valid large pool allocation, checking large session pool...
*9d619000 : large page allocation, Tag is Adbe, size is 0x4018 bytes
        Pooltag Adbe : Adobe's font driver

---

The surrounding code is a loop over a linked list of structures; the ESI 
register at the time of the crash holds a [F/B]link address which points to 
unmapped memory, potentially suggesting that the root cause of the crash is a 
use-after-free condition, or some kind of corruption of the linked list. Since 
the invalid address is later used to manipulate memory, we expect that this 
issue could be used to achieve remote code execution in the security context of 
the Windows kernel.

Moreover, we have also encountered similar crashes one instruction further in 
the code - in these cases, ESI turned out to accidentally point to mapped 
memory, but its contents did not correspond to the ATMFD's expectations 
(reinforcing the use-after-free hypothesis), thus resulting in an attempt to 
dereference a completely wild memory address, e.g.:

---
PAGE_FAULT_IN_NONPAGED_AREA (50)
Invalid system memory was referenced.  This cannot be protected by try-except,
it must be protected by a Probe.  Typically the address is just plain bad or it
is pointing at freed memory.
Arguments:
Arg1: eebd8451, memory referenced.
Arg2: 00000000, value 0 = read operation, 1 = write operation.
Arg3: 9205440e, If non-zero, the instruction address which referenced the bad 
memory
    address.
Arg4: 00000002, (reserved)

---

The issue reproduces on Windows 7 and 8.1. It is easiest to reproduce with 
Special Pools enabled for ATMFD.DLL (leading to an immediate crash when the bug 
is triggered), but it should also be possible to observe a crash on a default 
Windows installation in ATMFD.DLL.

Note that this crash is very similar in its nature to Issue 383 (invalid memory 
access while traversing a linked list of "Adbe" structures); however, due to 
uncertainty about the reason of the crash and the different stack traces, I am 
filing this as a separate report for now.

Attached is an archive with six proof of concept font files (three for each 
crash location) together with corresponding kernel crash logs.

This bug is subject to a 90 day disclosure deadline. If 90 days elapse without 
a broadly available patch, then the bug report will automatically become 
visible to the public.

Original issue reported on code.google.com by mjurc...@google.com on 15 May 2015 at 1:44

Attachments:

GoogleCodeExporter commented 8 years ago

Original comment by mjurc...@google.com on 21 May 2015 at 12:49

GoogleCodeExporter commented 8 years ago

Original comment by mjurc...@google.com on 22 May 2015 at 10:08

GoogleCodeExporter commented 8 years ago
Issue 393 has been merged into this issue.

Original comment by mjurc...@google.com on 9 Jul 2015 at 4:23

GoogleCodeExporter commented 8 years ago
Fixed in https://technet.microsoft.com/library/security/MS15-080.

Original comment by mjurc...@google.com on 11 Aug 2015 at 8:48

GoogleCodeExporter commented 8 years ago

Original comment by mjurc...@google.com on 12 Aug 2015 at 12:07

GoogleCodeExporter commented 8 years ago

Original comment by mjurc...@google.com on 18 Aug 2015 at 11:14