Awesome

AirSpy

Tool for exploring Apple's AirDrop protocol implementation on i/macOS, from the server's perspective.

Dumps requests and responses along with a linear code coverage trace of the code processing each request.

Building

$ git clone https://github.com/nowsecure/airspy.git
$ cd airspy/
$ npm install

Running

To spy on the implementation:

# On a local macOS system:
$ node dist/bin/airspy.js

# Or on a USB-connected iOS device:
$ node dist/bin/airspy.js -U

Then pop open the AirDrop UI on a nearby i/macOS device. This should result in data being captured and written to out/$serial/events.log, where $serial is a zero-based number incrementing with each run. Each request/response also gets written out to separate files for easy inspection and diffing.

It is also possible to replay an events.log from a previous run, which will re-generate the other output files:

$ node dist/bin/airspy.js -r out/0/events.log

This is also useful if you want to tweak the parsing of the requests to generate better or additional output artifacts. (PRs welcome!)

Sample output directory:

$ ls -1 out/0/
001-post-discover-coverage-modules.log
001-post-discover-coverage-symbols.log
001-post-discover-request-body.plist
001-post-discover-request-head.txt
001-post-discover-response-body.plist
001-post-discover-response-head.txt
002-post-ask-coverage-modules.log
002-post-ask-coverage-symbols.log
002-post-ask-request-body.plist
002-post-ask-request-head.txt
002-post-ask-response-body.plist
002-post-ask-response-head.txt
003-post-upload-request-head.txt
003-post-upload-response-head.txt
events.log
$

Then you may want to compare the code coverage traces for two requests.

For example to compare the modules involved, and in which order:

$ diff -u 001-post-discover-coverage-modules.log 002-post-ask-coverage-modules.log

And to compare the basic blocks involved, and in which order:

$ diff -u 001-post-discover-coverage-symbols.log 002-post-ask-coverage-symbols.log

One example is that by looking at where execution first diverges, you immediately know where the implementation decides what kind of request it's dealing with, so you can inspect that code with r2. Or, you might want to use it to guide a fuzzer.

Development workflow

To continuously recompile on change, keep this running in a terminal:

$ npm run build-app:watch

Plus another terminal with:

$ npm run build-agent:watch

And use an editor like Visual Studio Code for code completion and instant type-checking feedback.