diff options
author | Tobias Thierer <tobiast@google.com> | 2017-02-09 18:34:54 +0000 |
---|---|---|
committer | Tobias Thierer <tobiast@google.com> | 2017-02-09 20:06:28 +0000 |
commit | 8ee6debb115ba7479dd166fe5eba95b38e0d61cb (patch) | |
tree | ba7c30fe40b4db07ffaa2ae1abd5250d1e7a4823 /annotations/generate_annotated_java_files.py | |
parent | be473851645e6c67e7e94c262c74838e73b09d8a (diff) |
Speed up DHPublicKeyTest; make DHPrivateKeyTest consistent.
DHPublicKeyTest.test_getParams() was generating 1024bit Diffie-Hellman
parameters; this operation is very slow and very variable, often taking
more than the 5 minutes allowed before the test times out.
In practice, parameters are not generated at random, especially
not on a mobile device; instead, in practice well-known parameters
are picked rather than generating a random one.
To make this the test both more realistic and faster to run, this
CL splits up DHPublicKeyTest.test_getParams() into two tests:
1.) generate a 192bit, rather than 1024bit, parameters
2.) use hard coded 2048bit parameters to generate keys.
The 2048-bit group from RFC 7919, Appendix A, was used.
The same is done to DHPrivateKeyTest.test_getParams(), which was
previously only doing 1.).
Historical context:
DHPrivateKeyTest historically had observed the same problems;
therefore, commit 8bc378b1cec65bc06766a14a9cc575fec931b418 in
Feb 2011 changed it to only use 192bit, i.e. do 1.);
DHPublicKeyTest was not touched at that time.
DHPublicKeyTest's timeout was increased in June 2015 (cts commit
d16b864e9537e172436bfb8a5de76e9e49fcea35) but that broke with
the switch to CTSv2.
Test: Check that DHPrivateKeyTest and DHPublicKeyTest pass
within a few seconds on a Nexus 6P.
Bug: 30741591
Change-Id: I6af0a70761557e4f4c1641cad7807d2211553e25
Diffstat (limited to 'annotations/generate_annotated_java_files.py')
0 files changed, 0 insertions, 0 deletions