Linki
- ,,(...)człowiek nigdy nie wyrobi sobie o nikim właściwego pojęcia .Stwarza obraz i kontent.
- Intruzi Michael Marshall E-BOOK, Fantastyka, fantasy
- Insight Intermediate Student's Book Podręcznik dla szkół ponadgimnazjalnych Wildman Jayne, Podręczniki, lektury
- Instrukcja montarzu subwofera xc90 instruction VCC-138791-1 (1), E-Book's
- Inteligencja emocjonalna. Poradnik dla rodziców Beatriz Serrano Garrido e-book, Poradniki
- Ingarden - Wstep do Fenomenologii Husserla wyklad 1, e-book, Filozofia, Ingarden Roman
- Informatyka SP KL 4-6. Podręcznik multimedialny pendrive + zbiór zadań. Zajęcia komputerowe. Lubię to! 2012 Kęska Michał e-book, Inne
- Informatyka Europejczyka. Program nauczania do zajęć komputerowych w szkole podstawowej w edukacji w Danuta Kiałka e-book, Podręczniki, lektury
- Indie. Mapa Marco Polo w skali 12 500 000 praca zbiorowa E-BOOK, Turystyka, mapy, atlasy
- Instrukcja stosowania kas rejestrujących z wzorcową dokumentacją i nowymi ewidencjami - Grzegorz Tomala e-book, B jak Biznes i ekonomia
- Informator Płacowy Wskaźniki I Stawki Aktualne Od 1 Stycznia 2015 R - Praca zbiorowa e-book, B jak Biznes i ekonomia
- zanotowane.pl
- doc.pisz.pl
- pdf.pisz.pl
- kolejkielecka.xlx.pl
|
[ Pobierz całość w formacie PDF ] J Comput Virol (2006) 2:187–210 DOI 10.1007/s11416-006-0020-2 ORIGINAL PAPER In-depth analysis of the viral threats with OpenOffice.org documents David de Drézigué · Jean-Paul Fizaine · Nils Hansma Received: 10 April 2006 / Revised: 20 June 2006 / Accepted: 4 July 2006 / Published online: 1 August 2006 © Springer-Verlag France 2006 Abstract This paper presents an in-depth analysis of the OpenOffice suite (release 2.0.3) with respect to viral threats, independently of any software flaw or vulner- ability. First we will identify, then analyse the differ- ent potential viral vectors of OpenOffice.org v2.0.3. Our examination applies to win32 and Unix-like platforms. For each identified vector, a detailed study will show to which extend the infection is possible. From then on we will define the solutions in order to maximize Open- Office.org security in the production field as well as in the office tools at the administration level. a subject liable to concerns. On top of issues related to information leaks [1], the risk linked to macro-virus has been a concern to the end-users [3] since the Con- cept macro-virus in 1995. The presence of macros in an office document is henceforth identified by the risk – at least potential – of macro-virus. Even though the antiviral offer enable to fight efficiently against the well- known macro-viruses or the identified techniques, it still remains pretty easy to conceive an undetected macro- virus by the current antivirus. The development and the release of OpenOffice free software suite [11] for a few years seemed to inspire new hopes – and maybe hypes – as far as security linked to office documents was concerned. The adoption of this suite by several countries and by foreign armies in particular (Singapore in 2006, the French Gendarmerie in 2005), some institutions, some universities… [12] led the media to see a sign of greater security into it, furthermore it costs almost nothing. The certainty of a greater security towards the viral risk has become an overspread feeling in the freeware commu- nity as much as in the mind of several end-users. The consequence is that the latters forget the security behav- iour and reaction – and more particularly the caution against the presence of macros – when opening office documents. What should one think about the security of OpenOffice suite regarding macro-viruses? Myth or reality? The OpenOffice suite, just like its commercial equiv- alent, contains an inserted development environment around several programming languages. The language OoBasic – the best known – has replaced the Visual Basic for Applications for macros writing. But other programming languages, more powerful than the plain OoBasic allow more sophisticated developments than Keywords Malware · Macro virus · Document malware · Self-reproduction · Security Policy · Antiviral Policy 1 Introduction For many years, the software sector of the office suite has been dominated by the supremacy of the Micro- soft Office suite. This suite has succeeded in offering the end-user a powerful and ergonomic environment. Nev- ertheless, the security of the products making up this suite ( Word text processing, Excel spreadsheet, presen- tation software such as PowerPoint …) has long been B D. de Drézigué N. Hansma Ecole Supérieure et d’Application des Transmissions, Laboratoire de virologie et de cryptologie, BP 18, 35998 Rennes Armées, France e-mail: labo.virologie@esat.terre.defense.gouv.fr D. de Drézigué · J. -P. Fi za i ne ( ) · N. Hansma Ecole des Systèmes de Combat et Armes Navals, Centre d’Instruction Naval, BP 500, 83800 Toulon Naval, France · 188 D. de Drézigué et al. the plain macro writing, no matter how complex they may be. These different kinds of execution contexts seem to argue in favour of the possible uses of macro- viruses for OpenOffice. Such a possibility has been previously evoked by Rautiainen [14] but this author considered only a few security aspects. Moreover, Rau- tiainen’s paper dealt with the OpenOffice version 1.0.x. The OpenOffice release 2.0.x contains many more pow- erful capabilities as far as infectious threats are con- cerned. This article introduces the real security analysis of the OpenOffice suite with regards to viral threat is con- cerned. Independently of any software vulnerability, the results show that not only can this suite not be consid- ered immune tomacro-viruses but above all, considering the current development condition of this suite, the later shows a higher rate of hazard than for her commercial competitor. In other words, any hazardous behaviour from the end-users may lead to serious outcomes over information security system. This security analysis has been recognized by the design and the testing of several operational 1 viral strains demonstrating that this risk is truly real and is giving cause for concern. Therefore any security policy should seriously consider the viral risk linked to this suite. This article is partly based on [2] and is organized as follow. Sect. 2 will present the main identified vectors of potential infections for this suite, through a step by step analytic approach. Sect. 3 will deal with the operat- ing of these infection vectors through malevolent codes. Finally, Sect. 4 will, on the one hand, ponder over the viral risk attached to the OpenOffice suite, and on the second hand, will introduce the algorithmic of some few viral stocks that have been developed and tested for validation purposes. Disclaimer: Developing proof-of-concepts is not a goal in itself. It remains the only scientific way to prove if a viral risk exists or not. Deficiency of antivirus may result in human casualties or even deaths, lost of jobs... when considering critical or very critical systems. Consequently any serious protection cannot rely only on the deployment of an antivirus, whatever may be its efficiency. No serious security policy of such critical systems could accept secu- rity enforcement without a proactive research including proof-of-concept developments. This study has been per- formed in the strict respect of the different existing laws. The source codes which have been developed will neither be disclosed nor published. Only reknowned and strictly identified IT security professionals are eligible for freely – and without any compensation of any kind – accessing these source code provided that they fill an application form to the Virology and Cryptology Laboratory of the Army Signals Academy. 2 Identification of the potential infectious vectors: OpenOffice.org V2.0.x analysis First we will analyze the file format in order to know its structure as well as its organization; which enable us to infer the way we will be able to manipulate it. Sec- ondly, we will analyze the installation, the configuration so as to identify the potential propagation vectors for all the well known viral infection technologies [4]. We will define a potential propagation vector as follows: – an automatic, scheduled or event-related execution point; – use of a regular human task; – execution transfer, or process creation; – weak environment identification: configuration of the suitable application for the malevolent code execution. 2.1 The OpenDocument V1.0 file format This file format is a standard introduced byOasis-Open, 2 in 2005. It has recently adopted by ISO/IEC in May 2006, and called ISO/IEC DIS 26300, on the demand of the European Commission [8] with the agreement of SunMicrosystems and Oasis-Open. This is a file for- mat based on the XML technology. It can be used with all kind of tools handling this technology. OpenDoc- ument supports various other kind of document like database, diagrams, drawings, slideshows and spread- sheets. This file format is also supported by a num- ber of various applications: AbiWord, Knomos, Koffice, OpenOffice, Scribus StartOffice and recently by Ibm Workspace but not with Office Microsoft. State of Mas- sachusetts has first adopted this format, and recently Europe as standard exchange document format. The structure and organization of the file depend on the pat- tern specified by Oasis Open 3 [9]. We are presenting its main features that show specific relevance in the chosen context. 2 http://www.oasis-open.org 3 A pattern is a document that describes the creation of an XML file format. 1 These strains have been developed for the Writer component but are easily transposable to other software components of the suite. In-depth analysis of the viral threats with OpenOffice.org documents 189 2.1.1 Analysis of a blank document It is regarding the XML files format that the official liter- ature [9, Chap. 2] provides a very useful understanding of an XML OpenOffice.org document structure. The “file” command on our blank reference file, points out that the file is a ZIP format archive. We notice that the archive is neither compressed nor protected by a password. This will not prevent an archive infection. The extraction of the archive is usually carried out with the help of any utility compatible with the ZIP format. Thus here follows the content of our blank reference file: 2.1.2 Analysis of a user document The reference test file is a simple document which con- tains styles, a rather important number of pages and its own text layout information. At this stage, neither macros, nor other objects (OLE objects, links, sound or video data...) or images are present yet. We consider again the previous analysis protocol. Let us compare this present user document with the blank reference one. We find again the same files and directories. Only the size of files content.xml and setting.xml have changed. Whereas the reference file has a size of 147,332 bytes, the user file has a size of 147,470 bytes. However the corresponding archives differ from one another in size of only five bytes. Moreover, the content of setting.xml file has changed. To make things clear, the content of the user document archive here follows: The META-INF directory contains a manifest XML format file [9, Chap. 17]. The latest classifies the paths towards other XML files of the archive as well as infor- mation describing the archive as the encoding algorithm along with its parameters, the checksumpasswords algo- rithm as well as the checksum value. The directories “ Configuration2 ” and “ Pictures ” are empty. By lexico- graphic order: – content.xml : this file is common to all types of Open- Office.org documents. It can contain the following items: scripts, font settings, automatic style and doc- ument body. – meta.xml : this file contains the document meta- information (author, date of the last action…), Let us now compare both archive sizes: – styles.xml : specifies the style used for the document, – setting.xml : points to the configuration such as the programapplicationwindow size or to printing infor- mation. 190 D. de Drézigué et al. 2.1.3 Comparison to a document with macros In a macro-virus context, the size of such a virus is lia- ble to have a rather large size ranging froma few bytes to hundreds of kilobytes. Its final size will greatly depend on its sophisticated level and of the script language that has been used to implement it. Let us suppose that the viral code has an average size of 10 Kb. Let us suppose in addition that the resulting archive (that of an infected file) has increased of only one Kb for every 500 added bytes of virus. If we consider, in a first approach, that the size linearly increases, 4 thenwe notice that consequently the archive’s size has increased of 20 Kb. From an oper- ational point of view, this size increase is rather limited, not to say negligible. This feature enables the design of very large-sized viral codes for OpenOffice.org whithout making the final size of the archive explode. This fact is worth considering as far as viral stealth features are considered, especially when infecting large-sized docu- ments. Let us now insert a macro in the user document we have just considered. The name of this macro is dicOOo and belongs to the standard macros library (installation of a dictionary). We notice that a new directory has been created (output extract produced when unzipping an archive): 2.1.4 Storage structure of macros in libraries This new directory contains the whole organisation of macros into directories. The new file in this file sub-tree corresponds to the new macro itself. The manifest.xml file has been modified in order to declare the presence of any macro and to setup the document accordingly. The set of all macro pathes have been added. Libraries of macros are stored in the Basic directory. A library is materialized by a sub-directory which has the same name as the library itself. This sub-directory con- tains the macros which are linked to this library. More- over, a library entry is created in the “script-lc.xml file and macros themselves are listed in the script-lb.xml file. During this in-depth study, we easily manage to add, erase or modify one or more libraries by means of sim- ple command lines. When infecting macros, any efficient malware attack must modify the files script-lc.xml and script-lb.xml , in order to avoid error on document open- ing. By considering this, we successfully manage to << play >> with the general structure of OpenOffice.org documents, without prompting any alert. The probably most surprising issue comes from the fact that any absent macro (when deleted for instance) does not cause any er- ror when opening the document. When editing macros, the deleted macro is listed but cannot be executed since it does not contain any code. OpenOffice considers only the content of the files script-lc.xml and script-lb.xml and does not make any consistency checking. Here follows the corresponding dumps (normal li- brary). The code of any macro is located between the two fol- lowing XML tags. Here are located all the informations required to per- form an infection of any macro. Let us notice that it is possible to change the macro language. An efficient attack will usefully consider this characteristic. The size of content.xml and manifest.xml have increased whereas for the two relevant archives (with and without macro), a significant increase of size has been noticed (1,779 bytes). This corresponds to the addi- tional meta-information which result from the macro creation. – Here is the Basic directory content of an Open- Office.org document: 4 The experiments we have conducted have shown that this prop- erty did hold most of the time. In-depth analysis of the viral threats with OpenOffice.org documents 191 Let us analyse now a protected document with a macro. In some frequent cases, we have noticed some worrying algorithmic problems, as far as password pro- tection is concerned. 5 In these cases, the listing of the rel- evant archive is the same before and after the protection. Let us have a more particular look at the macro direc- tory. We notice a very important fact: there is strictly no difference at all between the corresponding macro files (documents with and without macros). Consequently, we can deduce that, in these cases, despite the pro- tection through passwords, the code of macros remains unprotected (unencrypted). This implies – and we will later confirm it with the proof-of-concepts codes we have developed – that infecting a password protected document is as easy as for unprotected ones. The integ- rity of some files in an archive is not taken into account, even by password protection. We manage to replace an encrypted macro with an unencrypted one. For that purpose, we modify the following data: META- INF/manifest.xml, Basic/Standard/ – Here is listed below the script-lc.xml content (two macro libraries): > and Basic/Standard/script-lb.xml. The password is still required at document opening while the new macro is executedwithout any security warning – by using trusted macros (see Sect 4.2). The analysis of the META-INF/manifest.xml enables to get some information about the OpenOffice encryp- tion which is moreover described in full in [9]. < macro_name.xml – Content of the script-lb.xml file (information about the content of the Standard library): – Only the following files have been encrypted: con- tent.xml, style.xml and setting.xml . – The encryption algorithm is Blowfish [17] in CFB mode. The encryption uses a different seed and ini- tialisation vector (IV) for every different file to pro- tect. Files are compressed before encryption. The initialisation vector is an arbitrary 8-bytes sequence which is base-64 encoded. – Encryption key is built directly from the password which is provided by the user. The key setup algo- rithm is the PBKDF2 algorithm [16] (16-bytes salt, base-64 encoded, 1024 iterations). Finally, the de- rived key is prepended to the encrypted text. – The document integrity is secured by means of the SHA-1 hash function [7] but there is no integrity checking for the macros which thus can be modified without prompting an alert to the user. Let us recall that we experiment the fact that it is possible to add a macro or even a whole library into an OpenOffice.org document without prompting any alert. A very interesting attack scenario then consists in call- ing or manipulating any newly added (by a malware) macro directly from another infected macro. Moreover, it is possible to load a library that OpenOffice cannot access directly [10, p. 764]. 2.1.5 Password protection Let us now suppose that the user has protected his docu- ment with a password (read and/or writemode). The rel- evant OpenOffice.org component will deny any access to the document unless the correct password is given upon request. Comparison of the content.xml files (for protected and unprotected documents) shows an unquestionable difference. However, the archive content is neither en- crypted nor protected by the password that has been used when saving the document. Only the content.xml file is encrypted. 5 We will not explicit these cases in order to limit their potential exploitation by attackers. We have contacted OpenOffice devel- opers in order to correct these security problems and a joint work has been initiated. We have noticed a very great reactivity.
[ Pobierz całość w formacie PDF ]
zanotowane.pldoc.pisz.plpdf.pisz.plzolka.keep.pl
|