Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
244 lines (174 sloc) 16.6 KB

iOS Network APIs

Almost every iOS app acts as a client to one or more remote services. As this network communication usually takes place over untrusted networks such as public Wifi, classical network based-attacks become a potential issue.

Most modern mobile apps use variants of HTTP based web-services, as these protocols are well-documented and supported. On iOS, the NSURLConnection class provides methods to load URL requests asynchronously and synchronously.

App Transport Security


App Transport Security (ATS) is a set of security checks that the operating system enforces when making connections with NSURLConnectionNSURLSession and CFURL to public hostnames. ATS is enabled by default for applications build on iOS SDK 9 and above.

ATS is enforced only when making connections to public hostnames. Therefore any connection made to an IP address, unqualified domain names or TLD of .local is not protected with ATS.

The following is a summarized list of App Transport Security Requirements:

  • No HTTP connections are allowed
  • The X.509 Certificate has a SHA256 fingerprint and must be signed with at least a 2048-bit RSA key or a 256-bit Elliptic-Curve Cryptography (ECC) key.
  • Transport Layer Security (TLS) version must be 1.2 or above and must support Perfect Forward Secrecy (PFS) through Elliptic Curve Diffie-Hellman Ephemeral (ECDHE) key exchange and AES-128 or AES-256 symmetric ciphers.

The cipher suite must be one of the following:

ATS Exceptions

ATS restrictions can be disabled by configuring exceptions in the Info.plist file under the NSAppTransportSecurity key. These exceptions can be applied to:

  • allow insecure connections (HTTP),
  • lower the minimum TLS version,
  • disable PFS or
  • allow connections to local domains.

ATS exceptions can be applied globally or per domain basis. The application can globally disable ATS, but opt in for individual domains. The following listing from Apple Developer documentation shows the structure of the [NSAppTransportSecurity]( "API Reference NSAppTransportSecurity") dictionary.

NSAppTransportSecurity : Dictionary {
    NSAllowsArbitraryLoads : Boolean
    NSAllowsArbitraryLoadsForMedia : Boolean
    NSAllowsArbitraryLoadsInWebContent : Boolean
    NSAllowsLocalNetworking : Boolean
    NSExceptionDomains : Dictionary {
        <domain-name-string> : Dictionary {
            NSIncludesSubdomains : Boolean
            NSExceptionAllowsInsecureHTTPLoads : Boolean
            NSExceptionMinimumTLSVersion : String
            NSExceptionRequiresForwardSecrecy : Boolean   // Default value is YES
            NSRequiresCertificateTransparency : Boolean

Source: Apple Developer Documentation.

The following table summarizes the global ATS exceptions. For more information about these exceptions, please refer to table 2 in the official Apple developer documentation.

Key Description
NSAllowsArbitraryLoads Disable ATS restrictions globally excepts for individual domains specified under NSExceptionDomains
NSAllowsArbitraryLoadsInWebContent Disable ATS restrictions for all the connections made from web views
NSAllowsLocalNetworking Allow connection to unqualified domain names and .local domains
NSAllowsArbitraryLoadsForMedia Disable all ATS restrictions for media loaded through the AV Foundations framework

The following table summarizes the per-domain ATS exceptions. For more information about these exceptions, please refer to table 3 in the official Apple developer documentation.

Key Description
NSIncludesSubdomains Indicates whether ATS exceptions should apply to subdomains of the named domain
NSExceptionAllowsInsecureHTTPLoads Allows HTTP connections to the named domain, but does not affect TLS requirements
NSExceptionMinimumTLSVersion Allows connections to servers with TLS versions less than 1.2
NSExceptionRequiresForwardSecrecy Disable perfect forward secrecy (PFS)

Starting from January 1 2017, Apple App Store review requires justification if one of the following ATS exceptions are defined.

  • NSAllowsArbitraryLoads
  • NSAllowsArbitraryLoadsForMedia
  • NSAllowsArbitraryLoadsInWebContent
  • NSExceptionAllowsInsecureHTTPLoads
  • NSExceptionMinimumTLSVersion

However this decline is extended later by Apple stating “To give you additional time to prepare, this deadline has been extended and we will provide another update when a new deadline is confirmed”

Analyzing the ATS Configuration

If the source code is available, open then Info.plist file in the application bundle directory and look for any exceptions that the application developer has configured. This file should be examined taking the applications context into consideration.

The following listing is an example of an exception configured to disable ATS restrictions globally.


If the source code is not available, then the Info.plist file should be either obtained from a jailbroken device or by extracting the application IPA file.

Since IPA files are ZIP archives, they can be extracted using any zip utility.

$ unzip app-name.ipa

Info.plist file can be found in the Payload/ directory of the extract. It’s a binary encoded file and has to be converted to a human readable format for the analysis.

plutil is a tool that’s designed for this purpose. It comes natively with Mac OS 10.2 and above versions.

The following command shows how to convert the Info.plist file into XML format.

$ plutil -convert xml1 Info.plist

Once the file is converted to a human readable format, the exceptions can be analyzed. The application may have ATS exceptions defined to allow it’s normal functionality. For an example, the Firefox iOS application has ATS disabled globally. This exception is acceptable because otherwise the application would not be able to connect to any HTTP website that does not have all the ATS requirements.

In general it can be summarised:

  • ATS should be configured according to best practices by Apple and only be deactivated under certain circumstances.
  • If the application connects to a defined number of domains that the application owner controls, then configure the servers to support the ATS requirements and opt-in for the ATS requirements within the app. In the following example, is owned by the application owner and ATS is enabled for that domain.
  • If connections to 3rd party domains are made (that are not under control of the app owner) it should be evaluated what ATS settings are not supported by the 3rd party domain and if they can be deactivated.
  • If the application opens third party web sites in web views, then from iOS 10 onwards NSAllowsArbitraryLoadsInWebContent can be used to disable ATS restrictions for the content loaded in web views

Testing Custom Certificate Stores and Certificate Pinning


Certificate pinning is the process of associating the mobile app with a particular X509 certificate of a server, instead of accepting any certificate signed by a trusted certificate authority. A mobile app that stores ("pins") the server certificate or public key will subsequently only establish connections to the known server. By removing trust in external certificate authorities, the attack surface is reduced (after all, there are many known cases where certificate authorities have been compromised or tricked into issuing certificates to impostors).

The certificate can be pinned during development, or at the time the app first connects to the backend. In that case, the certificate associated or 'pinned' to the host at when it seen for the first time. This second variant is slightly less secure, as an attacker intercepting the initial connection could inject their own certificate.

Static Analysis

Verify that the server certificate is pinned. Pinning can be implemented in multiple ways:

  1. Including server's certificate in the application bundle and performing verification on each connection. This requires an update mechanisms whenever the certificate on the server is updated
  2. Limiting certificate issuer to e.g. one entity and bundling the intermediate CA's public key into the application. In this way we limit the attack surface and have a valid certificate.
  3. Owning and managing your own PKI. The application would contain the intermediate CA's public key. This avoids updating the application every time you change the certificate on the server, due to e.g. expiration. Note that using your own CA would cause the certificate to be self-singed.

The code presented below shows how it is possible to check if the certificate provided by the server matches the certificate stored in the app. The method below implements the connection authentication and tells the delegate that the connection will send a request for an authentication challenge.

The delegate must implement connection:canAuthenticateAgainstProtectionSpace: and connection: forAuthenticationChallenge. Within connection: forAuthenticationChallenge, the delegate must call SecTrustEvaluate to perform customary X509 checks. The snippet below implements a check of the certificate.

(void)connection:(NSURLConnection *)connection willSendRequestForAuthenticationChallenge:(NSURLAuthenticationChallenge *)challenge
  SecTrustRef serverTrust = challenge.protectionSpace.serverTrust;
  SecCertificateRef certificate = SecTrustGetCertificateAtIndex(serverTrust, 0);
  NSData *remoteCertificateData = CFBridgingRelease(SecCertificateCopyData(certificate));
  NSString *cerPath = [[NSBundle mainBundle] pathForResource:@"MyLocalCertificate" ofType:@"cer"];
  NSData *localCertData = [NSData dataWithContentsOfFile:cerPath];
  The control below can verify if the certificate received by the server is matching the one pinned in the client.
  if ([remoteCertificateData isEqualToData:localCertData]) {
  NSURLCredential *credential = [NSURLCredential credentialForTrust:serverTrust];
  [[challenge sender] useCredential:credential forAuthenticationChallenge:challenge];
else {
  [[challenge sender] cancelAuthenticationChallenge:challenge];

Dynamic Analysis

Server certificate validation

Our test approach is to gradually relax security of the SSL handshake negotiation and check which security mechanisms are enabled.

  1. Having Burp set up as a proxy, make sure that there is no certificate added to the trust store (Settings -> General -> Profiles) and that tools like SSL Kill Switch are deactivated. Launch your application and check if you can see the traffic in Burp. Any failures will be reported under 'Alerts' tab. If you can see the traffic, it means that there is no certificate validation performed at all. If however, you can't see any traffic and you have an information about SSL handshake failure, follow the next point.
  2. Now, install Burp certificate, as explained in the portswigger user documentation. If the handshake is successful and you can see the traffic in Burp, it means that certificate is validated against device's trust store, but the pinning is not performed.
  3. If executing instructions from previous step doesn't lead to traffic being proxied through burp, it means that certificate is actually pinned and all security measures are in place. However, you still need to bypass the pinning in order to test the application. Please refer to section "Basic Security Testing" for more information on this.
Client certificate validation

Some applications use two-way SSL handshake, meaning that application verifies server's certificate and server verifies client's certificate. You can notice this if there is an error in Burp 'Alerts' tab indicating that client failed to negotiate connection.

There is a couple of things worth noting:

  1. The client certificate contains a private key that will be used for the key exchange.
  2. Usually the certificate would also need a password to use (decrypt) it.
  3. The certificate can be stored in the binary itself, data directory or in the keychain.

Most common and improper way of doing two-way handshake is to store the client certificate within the application bundle and hardcode the password. This obviously does not bring much security, because all clients will share the same certificate.

Second way of storing the certificate (and possibly password) is to use the keychain. Upon first login, the application should download the personal certificate and store it securely in the keychain.

Sometimes applications have one certificate that is hardcoded and use it for the first login and then the personal certificate is downloaded. In this case, check if it's possible to still use the 'generic' certificate to connect to the server.

Once you have extracted the certificate from the application (e.g. using Cycript or Frida), add it as client certificate in Burp, and you will be able to intercept the traffic.


OWASP Mobile Top 10 2016
  • V5.1: "Data is encrypted on the network using TLS. The secure channel is used consistently throughout the app."
  • V5.2: "The TLS settings are in line with current best practices, or as close as possible if the mobile operating system does not support the recommended standards."
  • V5.3: "The app verifies the X.509 certificate of the remote endpoint when the secure channel is established. Only certificates signed by a trusted CA are accepted."
  • V5.4: "The app either uses its own certificate store, or pins the endpoint certificate or public key, and subsequently does not establish connections with endpoints that offer a different certificate or key, even if signed by a trusted CA."
  • CWE-319 - Cleartext Transmission of Sensitive Information
  • CWE-326 - Inadequate Encryption Strength
  • CWE-295 - Improper Certificate Validation