RE: Portcullis Security Advisory 05-011 ACPI 1.6 BIOS

From: Paul J Docherty (
Date: 03/31/05

  • Next message: Thierry Carrez: "[ GLSA 200503-36 ] netkit-telnetd: Buffer overflow"
    Date: Thu, 31 Mar 2005 11:57:32 +0100
    To: "Chris Paget" <>

    Perhaps you are! The MBR is not being corrupted, it is intact and
    perfectly in order (if an MBR RFC existed it would be fully compliant).
    Whilst I agree, with privileged access you can DoS a system in many
    ways, all the ones you note would be obvious to a user and can be
    recovered from using common sense and some diagnostics for example
    re-flash or replace BIOS (does work on many systems) re-install an OS,
    re-partition etc etc. Also, local access to a system does not
    necessarily mean you have privileged access, just because your in the
    same room as a computer doesn't mean you can access it as an Admin/root.

    The issue described here requires the USER to do the things you have
    mentioned to recover hence destroy their own data, re-partition the disk
    etc. Subtly different, but different.

    Further, this issue was used to make a hard disk encryption (sector
    based) system fail-over requiring an administrator to uninstall it and
    hence decrypt the data.

    As they say, "One man's bread is another man's poison."


    -----Original Message-----
    From: Chris Paget []
    Sent: Thursday, March 31, 2005 10:43 AM
    To: Paul J Docherty
    Cc: bugs; Bugtraq; secunia
    Subject: Re: Portcullis Security Advisory 05-011 ACPI 1.6 BIOS

    Maybe I'm missing something, but I fail to see how this can be classed
    as a vulnerability? In order to corrupt the MBR in this way you would
    need privileged access to the machine (since you will be writing to raw
    sectors of the hard drive); if you have this level of access you can DoS

    the machine in any number of ways:
    1) Flash the BIOS with /dev/random.
    2) Format the hard drive
    3) Delete the partition table
    4) Disable the "active" flag on the primary partition
    5) Modify the CPU microcode
    ...and the list goes on. This is to mention nothing of software-based
    attacks; an OS can be killed in many, many ways if you can write to any
    of the OS files.

    If you're after a subtle attack, I doubt many people would even think to

    check whether their primary partition was still active. If you're after

    an effective attack, flashing the BIOS is non-recoverable on most
    systems (requiring a new motherboard).

    Any of these require the same level of access to the system, and produce

    the same effects. How is your "attack" any different?


    Chris Paget
    Paul J Docherty wrote:
    > Portcullis Security Advisory
    > Vulnerable System: 
    > This vulnerability affects any workstation running the ACPI 1.6 BIOS
    > implementation. 
    > Vulnerability Title: 
    > BIOS code logic error 
    > Vulnerability discovery and development: 
    > The Portcullis R&D team discovered this vulnerability. Whilst
    > a secured (software access control) laptop environment the team
    > discovered it was possible to cause any system running the vulnerable
    > version of the BIOS to become unstable and no longer boot. Once the
    > coding error was known it could be exploited to cause a near permanent
    > Denial of Service attack. 
    > Affected systems: 
    > The fault was initially found on a Toshiba Satellite Pro A60 running
    > ACPI BIOS version 1.60. This is a fundamental error that could be
    > exploited in any workstation based on the 1.60 version of the ACPI
    > All hard disks, including those encrypted by software or hardware,
    > be affected. Note that later versions of the ACPI (v1.7 and v1.8),
    > appear to contain the same coding error however, Portcullis have not
    > tested these versions. 
    > Details:
    > There is a programming error in the BIOS when the Master Boot Record
    > (MBR) is searched for the bootable disk partition. The error in the
    > code results in only the first slot in the MBR partition table being
    > tested for the active partition. Where the active partition is not
    > described in the first slot of the MBR table, the BIOS ignores the
    > remaining slots and searches other devices for a boot mechanism, and
    > consequence fails to start the Operating System. 
    > The BIOS seeks to validate the contents of the MBR and the boot sector
    > in the active partition to achieve confidence in the integrity of the
    > boot mechanism. At various points on the start-up sequence the BIOS
    > reads the MBR from disk and tests the data therein. In at least one
    > case, the BIOS sets out to test all 4 entries in the MBR partition
    > but due to treating an absolute pointer as one relative to the start
    > the MBR, it drops out of the loop after the first iteration. Slots 2,
    > and 4 are never examined and the active partition not found.
    > Specifically, the 512 byte MBR end-of-table is tested by comparing the
    > pointer to see if it has reached end-of-table at 510 bytes.
    > Unfortunately, in the ACPI version the pointer is set to MBR-start
    > start-of-table within MBR. Therefore, the pointer begins with a value
    > larger than the size of the MBR. The result is that the first slot
    > fails, when testing for the active partition byte, the test will show
    > end-of-table result: the subsequent slots are ignored.
    > An example of the problem is shown when the MBR is read into boot
    > address 0000:7C00h and then tested for an active partition.
    >     xor bx, bx             ;zero BX
    >     mov es, bx
    >     mov bx, 7C00h ;buffer address ES:BX set to 0000:7C00h
    >     mov cx, 1 ;set sector and cylinder value
    >     xor dh, dh ;and head value to read cylinder 0, head 0, sector 1
    >     mov dl, 80h ;set first hard disk drive
    >     mov ax, 0201h ;set read 1 sector
    >     int 13h ;call BIOS routine to read MBR and assume success
    >     add bx, 446 ;ES:BX point to start of partition table in MBR (BX =
    > 7DBEh)
    > NextSlot:
    >     cmp es:[bx], 80h ;is this slot bootable?
    >     je ActiveOk ;yes, drop out of loop
    >     add bx, 16 ;increment by length of slot in table 
    > (BX becomes 7C00h+446+16)
    >     cmp bx, 510 ;end of slot table reached? error! 
    > should be cmp bx, 7C00h + 510
    >     jb NextSlot ;no, look at next entry - branch never taken
    >     stc ;set no active partition found status
    >     ret
    > ActiveOk:
    >     clc ;set active partition found status
    > ret
    > Impact:
    > The system will not boot. Standard analysis tools will not identify
    > contents of the MBR partition as invalid. Unless the help-desk
    > is aware of this BIOS error, he/she will be unable to diagnose the
    > fault. Therefore, the impact is denial of service: the system will not
    > boot and yet there is nothing wrong with the contents of the hard
    > Exploit:
    > The error in the BIOS code means that any workstation using this
    > of the BIOS, can be configured such that the bootable partition is
    > defined below the first slot in the MBR partition table and will not
    > boot. An attack at any time during an operating session can leave the
    > workstation subsequently unable to reboot. The nature of the fault
    > that it is very difficult to identify and may leave the workstation
    > inoperable for an extended period of time.
    > Vendor Notified:
    > Toshiba were notified initially by email, as no response was received
    > this was followed up by telephone on Friday 4-Feb-2005 and again on
    > Wednesday 23-Feb-2005.
    > Vendor Response:
    > The A60 laptop works with their supplied configuration. Fault not
    > recognised. Reference supplied by Toshiba is 1-585 92244. When the
    > was reported again on 23-Feb-2005 a new fault reference was supplied
    > 1-165 162202 but no technical response was provided by Toshiba.
    > Workaround/Fix
    > Either, the BIOS code must be corrected by the manufacturer, or a
    > monitoring device installed to detect any re-configuration that would
    > exploit the vulnerability. Validating and correcting the contents of
    > MBR at system shutdown is a minimum requirement in these
    > Copyright: 
    > Copyright (c) Portcullis Computer Security Limited 2005. All rights
    > reserved worldwide.
    > *************************************************************
    > The information in this email is confidential and may be
    > legally privileged. It is intended solely for the addressee.
    > Any opinions expressed are those of the individual and do not
    > represent the opinion of the organisation. 
    > Access to this email by persons other than the intended
    > recipient is strictly prohibited.
    > If you are not the intended recipient, any disclosure, copying,
    > distribution or other action taken or omitted to be taken in
    > reliance on it, is prohibited and may be unlawful. 
    > When addressed to our clients any opinions or advice contained
    > in this email is subject to the terms and conditions expressed
    > in the applicable Portcullis Computer Security Limited terms
    > of business.
    > **************************************************************

  • Next message: Thierry Carrez: "[ GLSA 200503-36 ] netkit-telnetd: Buffer overflow"

    Relevant Pages

    • RE: Portcullis Security Advisory 05-011 ACPI 1.6 BIOS
      ... Fdisk /MBR only replaces the boot code within the sector, ... change in any way the Partition Information Block. ... Portcullis Security Advisory 05-011 ACPI 1.6 BIOS ... zeroing out the MBR). ...
    • Portcullis Security Advisory 05-011 ACPI 1.6 BIOS
      ... This vulnerability affects any workstation running the ACPI 1.6 BIOS ... code results in only the first slot in the MBR partition table being ... Where the active partition is not ...
      ... The following security advisory is sent to the securiteam mailing list, and can be found at the SecuriTeam web site: ... A vulnerability in ACPI BIOS ... There is a programming error in the BIOS when the Master Boot Record (MBR) ... Where the active partition is not described in the ...
    • Re: VM oder "Parallelbetrieb"?
      ... auch "aktive Partition ist keine erweiterte Partition". ... Es sucht am Ende des MBR die aktive Partition. ... wenn das BIOS wirklich den Boot-Code im MBR zur Ausfuehrung ... was bei "uebergenauen Plausibilitaetspruefungen nicht ...
    • Re: Booting process
      ... It accepts cpu instruction execution from the BIOS. ... You have control of the cpu. ... you could boot your OS from the MBR. ... MBR sector and the first partition. ...