Looking inside the (Drop) box

Security Analysis of Dropbox.

Web-based Presentation

• http://www.openwall.com/presentations/WOOT13-Security-Analysis-of-Dropbox/

"Upstream" Resources

• https://www.usenix.org/looking-inside-drop-box

  (includes video now!)

Status

We are able to handle Dropbox version 73.4.118 from 29-May-2019.

Reversing Dropbox

0. Note: For handling modern (> late 2018) Dropbox versions use "dedrop-ng" which is included in this repository.

   https://github.com/kholia/dedrop/tree/master/src/dedrop-ng

1. Download Dropbox and extract it.

   $ cd ~
   
   $ export DROPBOX_VERSION="dropbox-lnx.x86_64-23.4.19"
   
   $ wget -c "https://www.dropbox.com/download?plat=lnx.x86_64" -O $DROPBOX_VERSION.tar.gz
   
   $ tar -xzf $DROPBOX_VERSION.tar.gz
   

2. Build "dedrop". Switch to this repository and do,

   $ cd src/dedrop
   
   $ make
   
   $ cp libdedrop.so ~
   

3. Use LD_PRELOAD and inject libdedrop.so into Dropbox.

   $ cd ~
   
   $ export BLOB_PATH=.dropbox-dist/$DROPBOX_VERSION/dropbox
   
   $ LD_PRELOAD=`pwd`/libdedrop.so .dropbox-dist/dropboxd
   

4. De-compile the "fixed" bytecode files.

   $ uncompyle6 pyc_decrypted/client_api/hashing.pyc
   ...
   

5. Study the soure-code, find bugs and make Dropbox better!

6. You might need to do xhost local:root to start Dropbox.

Dependencies (for paper)

• texlive

• texlive-minted

• texlive-texments

• python-pygments

  yum install texlive texlive-minted python-pygments texlive-texments \
      texlive-ifplatform texlive-endnotes
  

Credits

• ReflectiveDLLInjection is written by Stephen Fewer

  See https://github.com/stephenfewer/ReflectiveDLLInjection.git

• uncompyle2

  • https://github.com/wibiti/uncompyle2
  • https://github.com/Mysterie/uncompyle2

Resources

• https://github.com/rocky/python-uncompyle6
• https://github.com/MyNameIsMeerkat/pyREtic

TODO

• Find alternatives to "tray_login" method since it is going to be patched soon. This is now redundant since Dropbox client now uses 2FA properly.
• "While your submission was interesting, there has been other research on similar topics. There is nothing wrong with talking about the same topic more than once, especially one that has a large impact but if you are expanding on a topic, make sure to highlight how you are taking the research to a new level. Be clear with the review board about how what you are doing is extending the research." <= (apply this feedback to the paper and presentation).
• Looking deeper into the (Drop) box.
  • dump bytecode from memory (revive pyREtic).