Update unit tests for download.py, repository_lib.py, repository_tool.py, and updater.py.
Re-generate repository data so that compressed metadata may be tested.
Reject downloaded metadata as early as possible. The top-level roles were all downloaded as a group and then their
expiration inspected. All metadata provided by a mirror that has already expired is discarded immediately and the next mirror tried. The update process stops if a requested role cannot be successfully validated, or one of its parents.
[2014-04-29 02:00:32,308 UTC] [tuf.download] [INFO] [_download_file:745@download.py]
Downloading: http://localhost:8001/metadata/timestamp.json
[2014-04-29 02:00:32,324 UTC] [tuf.download] [INFO] [_check_downloaded_length:676@download.py]
Downloaded 544 bytes out of an upper limit of 16384 bytes.
[2014-04-29 02:00:32,324 UTC] [tuf.client.updater] [INFO] [_get_file:1189@updater.py]
Not decompressing http://localhost:8001/metadata/timestamp.json
[2014-04-29 02:00:32,331 UTC] [tuf.download] [INFO] [_download_file:745@download.py]
Downloading: http://localhost:8001/metadata/snapshot.json
[2014-04-29 02:00:32,333 UTC] [tuf.download] [INFO] [_check_downloaded_length:654@download.py]
Downloaded 1003 bytes out of the expected 1003 bytes.
[2014-04-29 02:00:32,334 UTC] [tuf.client.updater] [INFO] [_get_file:1189@updater.py]
Not decompressing http://localhost:8001/metadata/snapshot.json
[2014-04-29 02:00:32,334 UTC] [tuf.client.updater] [INFO] [_check_hashes:696@updater.py]
The file's sha256 hash is correct: 5b3aec7cf295a25e4b39d875c7474511da9645bc6d27f9e86fb7e439c82e0ec7
[2014-04-29 02:00:32,335 UTC] [tuf.client.updater] [ERROR] [_ensure_not_expired:1789@updater.py]
Metadata 'snapshot' expired on Tue Apr 29 01:59:01 2014 (UTC).
Do not request, download, and install top-level roles if the root of trust has already expired after the inital load. If requested, update an expired root role:
[2014-04-29 01:18:02,457 UTC] [tuf.client.updater] [ERROR] [_ensure_not_expired:1789@updater.py]
Metadata 'root' expired on Mon Apr 28 23:23:57 2014 (UTC).
[2014-04-29 01:18:02,458 UTC] [tuf.client.updater] [INFO] [refresh:628@updater.py]
Expired Root metadata was loaded from disk. Try to update it now.
[2014-04-29 01:18:02,458 UTC] [tuf.download] [INFO] [_download_file:745@download.py]
Downloading: http://localhost:8001/metadata/root.json
[2014-04-29 01:18:02,461 UTC] [tuf.download] [INFO] [_check_downloaded_length:676@download.py]
Downloaded 1198 bytes out of an upper limit of 512000 bytes.
[2014-04-29 01:18:02,461 UTC] [tuf.client.updater] [INFO] [_get_file:1189@updater.py]
Not decompressing http://localhost:8001/metadata/root.json
[2014-04-29 01:18:02,462 UTC] [tuf.client.updater] [ERROR] [_ensure_not_expired:1789@updater.py]
Metadata 'root' expired on Mon Apr 28 23:23:57 2014 (UTC).
Note: An expired 'root' was provided by the server. The requested root must also be signed by keys trusted by the client.
Ensure the target file used in the slow retrieval attack is larger than tuf.conf.SLOW_START_GRACE_PERIOD. The previous size of the test file might have led to inconsistent triggering of a slow retrieval error.
Update docstrings and comments and complete the initial implementation of issue 151.
Adjust logger level for tuf.download._check_downloaded_length().
Initial implementation of issue 137.
You may argue that the redundancy is unnecessary (pun intended), but it
is there because redundancy means one safety check will work where
another fails. I introduced this redundant file length check because the
updater unit test is mocking the download functions, which means that
file length checks in the download functions are being bypassed.
Redundancy is a good thing for safety.
download.py:Add a timeout and rewrite the _fileobject.read()
test_slow_retrieval_attack.py:Add a new kind of slow retrieval attack
slow_retrieval_server.py:Modification for new kind of slow retrieval attack
modification of updater.py for download.py
modification of conf.py for fix
modification of test_download.py for download.py
modification of test_updater.py for download.py
add a new test of endless data attack to metadata timestamp.txt
more readable and fix the endless data attack issue.
His code splits tuf.download.download_url_to_tempfileobj into two major
pieces. The first piece opens a connection to a URL, and computes
the required and reported lengths for downloading data from that given
URL. The second piece downloads data from the given URL in such a way
that we can defend against endless data and slow retrieval attacks.
Do not wish to risk breaking tuf.client.updater at this point,
where the fault lies because metadata metadata is not checked.
We are aware of #26 and #33, and actively working to fix them.
