EVGalex

lundi 13 avril 2020

PDFex: Major Security Flaws In PDF Encryption

After investigating the security of PDF signatures, we had a deeper look at PDF encryption. In cooperation with our friends from Münster University of Applied Sciences, we discovered severe weaknesses in the PDF encryption standard which lead to full plaintext exfiltration in an active-attacker scenario.

To guarantee confidentiality, PDF files can be encrypted. This enables the secure transfer and storing of sensitive documents without any further protection mechanisms.

The key management between the sender and recipient may be password based (the recipient must know the password used by the sender, or it must be transferred to them through a secure channel) or public key based (i.e., the sender knows the X.509 certificate of the recipient).

In this research, we analyze the security of encrypted PDF files and show how an attacker can exfiltrate the content without having the corresponding keys.

So what is the problem?

The security problems known as PDFex discovered by our research can be summarized as follows:

Even without knowing the corresponding password, the attacker possessing an encrypted PDF file can manipulate parts of it.
More precisely, the PDF specification allows the mixing of ciphertexts with plaintexts. In combination with further PDF features which allow the loading of external resources via HTTP, the attacker can run direct exfiltration attacks once a victim opens the file.
PDF encryption uses the Cipher Block Chaining (CBC) encryption mode with no integrity checks, which implies ciphertext malleability.
This allows us to create self-exfiltrating ciphertext parts using CBC malleability gadgets. We use this technique not only to modify existing plaintext but to construct entirely new encrypted objects.

Who uses PDF Encryption?

PDF encryption is widely used. Prominent companies like Canon and Samsung apply PDF encryption in document scanners to protect sensitive information.

Further providers like IBM offer PDF encryption services for PDF documents and other data (e.g., confidential images) by wrapping them into PDF. PDF encryption is also supported in different medical products to transfer health records, for example Innoport, Ricoh, Rimage.

Due to the shortcomings regarding the deployment and usability of S/MIME and OpenPGP email encryption, some organizations use special gateways to automatically encrypt email messages as encrypted PDF attachments, for example CipherMail, Encryptomatic, NoSpamProxy. The password to decrypt these PDFs can be transmitted over a second channel, such as a text message (i.e., SMS).

Technical details of the attacks

We developed two different attack classes on PDF Encryption: Direct Exfiltration and CBC Gadgets.

Attack 1: Direct Exfiltration (Attack A)

The idea of this attack is to abuse the partial encryption feature by modifying an encrypted PDF file. As soon as the file is opened and decrypted by the victim sensitive content is sent to the attacker. Encrpyted PDF files does not have integrity protection. Thus, an attacker can modify the structure of encrypted PDF documents, add unencrypted objects, or wrap encrypted parts into a context controlled the attacker.

In the given example, the attacker abuses the flexibility of the PDF encryption standard to define certain objects as unencrypted. The attacker modifies the Encrypt dictionary (6 0 obj) in a way that the document is partially encrypted – all streams are left AES256 encrypted while strings are defined as unencrypted by setting the Identity filter. Thus, the attacker can freely modify strings in the document and add additional objects containing unencrypted strings.

The content to be exfiltrated is left encrypted, see Contents (4 0 obj) and EmbeddedFile (5 0 obj). The most relevant object for the attack is the definition of an Action, which can submit a form, invoke a URL, or execute JavaScript. The Action references the encrypted parts as content to be included in requests and can thereby be used to exfiltrate their plaintext to an arbitrary URL. The execution of the Action can be triggered automatically once the PDF file is opened (after the decryption) or via user interaction, for example, by clicking within the document.

This attack has three requirements to be successful. While all requirements are PDF standard compliant, they have not necessarily been implemented by every PDF application:

Partial encryption: Partially encrypted documents based on Crypt Filters like the Identity filter or based on other less supported methods like the None encryption algorithm.
Cross-object references: It must be possible to reference and access encrypted string or stream objects from unencrypted attacker-controlled parts of the PDF document.
Exfiltration channel: One of the interactive features allowing the PDF reader to communicate via Internet must exist, with or without user interaction. Such Features are PDF Forms, Hyperlinks, or JavaScript.

Please note that the attack does not abuse any cryptographic issues, so that there are no requirements to the underlying encryption algorithm (e.g., AES) or the encryption mode (e.g., CBC).

In the following, we show three techniques how an attack can exfiltrate the content.

Exfiltration via PDF Forms (A1)

The PDF standard allows a document's encrypted streams or strings to be defined as values of a PDF form to be submitted to an external server. This can be done by referencing their object numbers as the values of the form fields within the Catalog object, as shown in the example on the left side. The value of the PDF form points to the encrypted data stored in 2 0 obj.

To make the form auto-submit itself once the document is opened and decrypted, an OpenAction can be applied. Note that the object which contains the URL (http://p.df) for form submission is not encrypted and completely controlled by the attacker. As a result, as soon as the victim opens the PDF file and decrypts it, the OpenAction will be executed by sending the decrypted content of 2 0 obj to (http://p.df).

Exfiltration via Hyperlinks (A2)

If forms are not supported by the PDF viewer, there is a second method to achieve direct exfiltration of a plaintext. The PDF standard allows setting a "base" URI in the Catalog object used to resolve all relative URIs in the document.

This enables an attacker to define the encrypted part as a relative URI to be leaked to the attacker's web server. Therefore the base URI will be prepended to each URI called within the PDF file. In the given example, we set the base URI to (http://p.df).

The plaintext can be leaked by clicking on a visible element such as a link, or without user interaction by defining a URI Action to be automatically performed once the document is opened.

In the given example, we define the base URI within an Object Stream, which allows objects of arbitrary type to be embedded within a stream. This construct is a standard compliant method to put unencrypted and encrypted strings within the same document. Note that for this attack variant, only strings can be exfiltrated due to the specification, but not streams; (relative) URIs must be of type string. However, fortunately (from an attacker's point of view), all encrypted streams in a PDF document can be re-written and defined as hex-encoded strings using the hexadecimal string notation.

Nevertheless, the attack has some notable drawbacks compared to Exfiltration via PDF Forms:

The attack is not silent. While forms are usually submitted in the background (by the PDF viewer itself), to open hyperlinks, most applications launch an external web browser.
Compared to HTTP POST, the length of HTTP GET requests, as invoked by hyperlinks, is limited to a certain size.
PDF viewers do not necessarily URL-encode binary strings, making it difficult to leak compressed data.

Exfiltration via JavaScript (A3)

The PDF JavaScript reference allows JavaScript code within a PDF document to directly access arbitrary string/stream objects within the document and leak them with functions such as *getDataObjectContents* or *getAnnots*.

In the given example, the stream object 7 is given a Name (x), which is used to reference and leak it with a JavaScript action that is automatically triggered once the document is opened. The attack has some advantages compared to Exfiltration via PDF Forms and Exfiltration via Hyperlinks, such as the flexibility of an actual programming language.

It must, however, be noted that – while JavaScript actions are part of the PDF specification – various PDF applications have limited JavaScript support or disable it by default (e.g., Perfect PDF Reader).

Attack 2: CBC Gadgets (Attack B)

Not all PDF viewers support partially encrypted documents, which makes them immune to direct exfiltration attacks. However, because PDF encryption generally defines no authenticated encryption, attackers may use CBC gadgets to exfiltrate plaintext. The basic idea is to modify the plaintext data directly within an encrypted object, for example, by prefixing it with an URL. The CBC gadget attack, thus does not necessarily require cross-object references.

Note that all gadget-based attacks modify existing encrypted content or create new content from CBC gadgets. This is possible due to the malleability property of the CBC encryption mode.

This attack has two necessary preconditions:

Known plaintext: To manipulate an encrypted object using CBC gadgets, a known plaintext segment is necessary. For AESV3 – the most recent encryption algorithm – this plain- text is always given by the Perms entry. For older versions, known plaintext from the object to be exfiltrated is necessary.
Exfiltration channel: One of the interactive features: PDF Forms or Hyperlinks.

These requirements differ from those of the direct exfiltration attacks, because the attacks are applied "through" the encryption layer and not outside of it.

Exfiltration via PDF Forms (B1)

As described above, PDF allows the submission of string and stream objects to a web server. This can be used in conjunction with CBC gadgets to leak the plaintext to an attacker-controlled server, even if partial encryption is not allowed.

A CBC gadget constructed from the known plaintext can be used as the submission URL, as shown in the example on the left side. The construction of this particular URL gadget is challenging. As PDF encryption uses PKCS#5 padding, constructing the URL using a single gadget from the known Perms plaintext is difficult, as the last 4 bytes that would need to contain the padding are unknown.

However, we identified two techniques to solve this. On the one hand, we can take the last block of an unknown ciphertext and append it to our constructed URL, essentially reusing the correct PKCS#5 padding of the unknown plaintext. Unfortunately, this would introduce 20 bytes of random data from the gadgeting process and up to 15 bytes of the unknown plaintext to the end of our URL.

On the other hand, the PDF standard allows the execution of multiple OpenActions in a document, allowing us to essentially guess the last padding byte of the Perms value. This is possible by iterating over all 256 possible values of the last plaintext byte to get 0x01, resulting in a URL with as little random as possible (3 bytes). As a limitation, if one of the 3 random bytes contains special characters, the form submission URL might break.

Exfiltration via Hyperlinks (B2)

Using CBC gadgets, encrypted plaintext can be prefixed with one or more chosen plaintext blocks. An attacker can construct URLs in the encrypted PDF document that contain the plaintext to exfiltrate. This attack is similar to the exfiltration hyperlink attack (A2). However, it does not require the setting of a "base" URI in plaintext to achieve exfiltration.

The same limitations described for direct exfiltration based on links (A2) apply. Additionally, the constructed URL contains random bytes from the gadgeting process, which may prevent the exfiltration in some cases.

Exfiltration via Half-Open Object Streams (B3)

While CBC gadgets are generally restricted to the block size of the underlying block cipher – and more specifically the length of the known plaintext, in this case, 12 bytes – longer chosen plaintexts can be constructed using compression. Deflate compression, which is available as a filter for PDF streams, allows writing both uncompressed and compressed segments into the same stream. The compressed segments can reference back to the uncompressed segments and achieve the repetition of byte strings from these segments. These backreferences allow us to construct longer continuous plaintext blocks than CBC gadgets would typically allow for. Naturally, the first uncompressed occurrence of a byte string still appears in the decompressed result. Additionally, if the compressed stream is constructed using gadgets, each gadget generates 20 random bytes that appear in the decompressed stream. A non-trivial obstacle is to keep the PDF viewer from interpreting these fragments in the decompressed stream. While hiding the fragments in comments is possible, PDF comments are single-line and are thus susceptible to newline characters in the random bytes. Therefore, in reality, the length of constructed compressed plaintexts is limited.

To deal with this caveat, an attacker can use ObjectStreams which allow the storage of arbitrary objects inside a stream. The attacker uses an object stream to define new objects using CBC gadgets. An object stream always starts with a header of space-separated integers which define the object number and the byte offset of the object inside the stream. The dictionary of an object stream contains the key First which defines the byte offset of the first object inside the stream. An attacker can use this value to create a comment of arbitrary size by setting it to the first byte after their comment.

Using compression has the additional advantage that compressed, encrypted plaintexts from the original document can be embedded into the modified object. As PDF applications often create compressed streams, these can be incorporated into the attacker-created compressed object and will therefore be decompressed by the PDF applications. This is a significant advantage over leaking the compressed plaintexts without decompression as the compressed bytes are often not URL-encoded correctly (or at all) by the PDF applications, leading to incomplete or incomprehensible plaintexts. However, due to the inner workings of the deflate algorithms, a complete compressed plaintext can only be prefixed with new segments, but not postfixed. Therefore, a string created using this technique cannot be terminated using a closing bracket, leading to a half-open string. This is not a standard compliant construction, and PDF viewers should not accept it. However, a majority of PDF viewers accept it anyway.

Evaluation

During our security analysis, we identified two standard compliant attack classes which break the confidentiality of encrypted PDF files. Our evaluation shows that among 27 widely-used PDF viewers, all of them are vulnerable to at least one of those attacks, including popular software such as Adobe Acrobat, Foxit Reader, Evince, Okular, Chrome, and Firefox.

You can find the detailed results of our evaluation here.

What is the root cause of the problem?

First, many data formats allow to encrypt only parts of the content (e.g., XML, S/MIME, PDF). This encryption flexibility is difficult to handle and allows an attacker to include their own content, which can lead to exfiltration channels.

Second, when it comes to encryption, AES-CBC – or encryption without integrity protection in general – is still widely supported. Even the latest PDF 2.0 specification released in 2017 still relies on it. This must be fixed in future PDF specifications and any other format encryption standard, without enabling backward compatibility that would re-enable CBC gadgets.

A positive example is JSON Web Encryption standard, which learned from the CBC attacks on XML and does not support any encryption algorithm without integrity protection.

Authors of this Post

Jens Müller
Fabian Ising
Vladislav Mladenov
Christian Mainka
Sebastian Schinzel
Jörg Schwenk

Acknowledgements

Many thanks to the CERT-Bund team for the great support during the responsible disclosure process.

Gridcoin - The Bad

In this post we will show why Gridcoin is insecure and probably will never achieve better security. Therefore, we are going to explain two critical implementation vulnerabilities and our experience with the core developer in the process of the responsible disclosure.

In our last blog post we described the Gridcoin architecture and the design vulnerability we found and fixed (the good). Now we come to the process of responsibly disclosing our findings and try to fix the two implementation vulnerabilities (the bad).

Update (15.08.2017):
After the talk at WOOT'17 serveral other developers of Gridcoin quickly reached out to us and told us that there was a change in responsibility internally in the Gridcoin-Dev team. Thus, we are going to wait for their response and then change this blog post accordingly. So stay tuned :)

Update (16.08.2017):
We are currently in touch with the whole dev team of Gridcoin and it seems that they are going to fix the vulnerabilities with the next release.

TL;DR
The whole Gridcoin currency is seriously insecure against attacks and should not be trusted anymore; unless some developers are in place, which have a profound background in protocol and application security.

What is Gridcoin?

Gridcoin is an altcoin, which is in active development since 2013. It claims to provide a high sustainability, as it has very low energy requirements in comparison to Bitcoin. It rewards users for contributing computation power to scientific projects, published on the BOINC project platform. Although Gridcoin is not as widespread as Bitcoin, its draft is very appealing as it attempts to eliminate Bitcoin's core problems. It possesses a market capitalization of $13,530,738 as of August the 4th 2017 and its users contributed approximately 5% of the total scientific BOINC work done before October 2016.

A detailed description of the Gridcoin architecture and technical terms used in this blog post are explained in our last blog post.

The Issues

Currently there are 2 implementation vulnerabilities in the source code, and we can mount the following attacks against Gridcoin:

We can steal the block creation reward from many Gridcoin minters
We can efficiently prevent many Gridcoin minters from claiming their block creation reward (DoS attack)

So why do we not just open up an issue online explaining the problems?

Because we already fixed a critical design issue in Gridcoin last year and tried to help them to fix the new issues. Unfortunately, they do not seem to have an interest in securing Gridcoin and thus leave us no other choice than fully disclosing the findings.

In order to explain the vulnerabilities we will take a look at the current Gridcoin source code (version 3.5.9.8).

WARNING: Due to the high number of source code lines in the source files, it can take a while until your browser shows the right line.

Stealing the BOINC block reward

The developer implemented our countermeasures in order to prevent our attack from the last blog post. Unfortunately, they did not look at their implementation from an attacker's perspective. Otherwise, they would have found out that they conduct not check, if the signature over the last block hash really is done over the last block hash. But we come to that in a minute. First lets take a look at the code flow:

In the figure the called-by-graph can be seen for the function VerifyCPIDSignature.

CheckBlock → DeserializeBoincBlock [Source]

Here we deserialize the BOINC data structure from the first transaction

CheckBlock → IsCPIDValidv2 [Source]

Then we call a function to verify the CPID used in the block. Due to the massive changes over the last years, there are 3 possible verify functions. We are interested in the last one (VerifyCPIDSignature), for the reason that it is the current verification function.

IsCPIDValidv2 → VerifyCPIDSignature [Source]
VerifyCPIDSignature → CheckMessageSignature [Source, Source]

In the last function the real signature verification is conducted [Source]. When we closely take a look at the function parameter, we see the message (std::string sMsg) and the signature (std::string sSig) variables, which are checked. But where does this values come from?

If we go backwards in the function call graph we see that in VerifyCPIDSignature the sMsg is the string sConcatMessage, which is a concatenation of the sCPID and the sBlockHash.
We are interested where the sBlockHash value comes from, due to the fact that this one is the only changing value in the signature generation.
When we go backwards, we see that the value originate from the deserialization of the BOINC structure (MiningCPID& mc) and is the variable mc.lastblockhash [Source, Source]. But wait a second, is this value ever checked whether it contains the real last block hash?

No, it is not....

So they just look if the stored values there end up in a valid signature.

Thus, we just need to wait for one valid block from a researcher and copy the signature, the last block hash value, the CPID and adjust every other dynamic value, like the RAC. Consequently, we are able to claim the reward of other BOINC users. This simple bug allows us again to steal the reward of every Gridcoin researcher, like there was never a countermeasure.

Lock out Gridcoin researcher
The following vulnerability allows an attacker under specific circumstances to register a key pair for a CPID, even if the CPID was previously tied to another key pair. Thus, the attacker locks out a legit researcher and prevent him from claiming BOINC reward in his minted blocks.

Reminder: A beacon is valid for 5 months, afterwards a new beacon must be sent with the same public key and CPID.

Therefore, we need to take a look at the functions, which process the beacon information. Every time there is a block, which contains beacon information, it is processed the following way (click image for higher resolution):

In the figure the called-by-graph can be seen for the function GetBeaconPublicKey.

We now show the source code path:

ProcessBlock → CheckBlock [Source]
CheckBlock → LoadAdminMessages [Source]
LoadAdminMessages → MemorizeMessages [Source]
MemorizeMessages → GetBeaconPublicKey [Source]

In the last function GetBeaconPublicKey there are different paths to process a beacon depending on the public key, the CPID, and the time since both were associated to each other.
For the following explanation we assume that we have an existing association (bound) between a CPID A and a public key pubK_A for 4 months.

First public key for a CPID received [Source]

The initial situation, when pubK_A was sent and bind to CPID A (4 months ago)

Existing public key for a CPID was sent [Source]

The case that pubK_A was resent for a CPID A, before the 5 months are passed by

Other public key for a CPID was sent [Source]

The case, if a different public key pubK_B for the CPID A was sent via beacon.

The existing public key for the CPID is expired

After 5 months a refresh for the association between A and pubK_A is required.

When an incoming beacon is processed, a look up is made, if there already exists a public key for the CPID used in the beacon. If yes, it is compared to the public key used in the beacon (case 2 and 3).
If no public key exists (case 1) the new public key is bound to the CPID.

If a public key exists, but it was not refreshed directly 12.960.000 seconds (5 months [Source]) after the last beacon advertisement of the public key and CPID, it is handled as no public key would exist [Source].

Thus, case 1 and 4 are treated identical, if the public key is expired, allowing an attacker to register his public key for an arbitrary CPID with expired public key. In practice this allows an attacker to lock out a Gridcoin user from the minting process of new blocks and further allows the attacker to claim reward for BOINC work he never did.

There is a countermeasure, which allows a user to delete his last beacon (identified by the CPID) . Therefore, the user sends 1 GRC to a special address (SAuJGrxn724SVmpYNxb8gsi3tDgnFhTES9) from an GRC address associated to this CPID [Source]. We did not look into this mechanism in more detail, because it only can be used to remove our attack beacon, but does not prevent the attack.

The responsible disclosure process

As part of our work as researchers we all have had the pleasure to responsible disclose the findings to developer or companies.

For the reasons that we wanted to give the developer some time to fix the design vulnerabilities, described in the last blog post, we did not issue a ticket at the Gridcoin Github project. Instead we contacted the developer at September the 14th 2016 via email and got a response one day later (2016/09/15). They proposed a variation of our countermeasure and dropped the signature in the advertising beacon, which would result in further security issues. We sent another email (2016/09/15) explained to them, why it is not wise to change our countermeasures and drop the signature in the advertising beacon.
Unfortunately, we did not receive a response. We tried it again on October the 31th 2016. They again did not respond, but we saw in the source code that they made some promising changes. Due to some other projects we did not look into the code until May 2017. At this point we found the two implementation vulnerabilities. We contacted the developer twice via email (5th and 16th of May 2017) again, but never received a response. Thus, we decided to wait for the WOOT notification to pass by and then fully disclose the findings. We thus have no other choice then to say that:

The whole Gridcoin cryptocurrency is seriously insecure against attacks and should not be trusted anymore; unless some developers are in place, which have a profound background in protocol and application security.

Further Reading

A more detailed description of the Gridcoin architecture, the old design issue and the fix will be presented at WOOT'17. Some days after the conference the paper will be available online.

C++ Std::Condition_Variable Null Pointer Derreference

This story is about a bug generated by g++ and clang compilers (at least)
The condition_variables is a feature on the standard library of c++ (libstdc++), when its compiled statically a weird asm code is generated.

Any example on the link below will crash if its compiled statically:
https://en.cppreference.com/w/cpp/thread/condition_variable

In this case the condition_variable.wait() crashed, but this happens with other methods, a simple way to trigger it:

If this program is compiled dynamically the crash doesn't occur:

Looking the dissasembly there is a surprise created by the compiler:

Compilers:
g++ 9.2.1+20200130-2
clang++ v9

Both compilers are generating the "call 0x00"

If we check this call in a dynamic compiled:

The implementation of condition_variable in github:
https://github.com/gcc-mirror/gcc/blob/b7c9bd36eaacac42631b882dc67a6f0db94de21c/libstdc%2B%2B-v3/include/std/condition_variable

The compilers can't copile well this code in static, and same happens on other condition_variable methods.
I would say the _lock is being assembled improperly in static, is not exacly a null pointer derreference but the effects are the same, executing code at address 0x00 which on linux is a crash on most of cases.