diff options
| author | Jack Lloyd <[email protected]> | 2016-10-13 14:36:28 -0400 |
|---|---|---|
| committer | Jack Lloyd <[email protected]> | 2016-10-13 14:36:28 -0400 |
| commit | 30c85ac981fdad46639ce02aa7e8612601fc5245 (patch) | |
| tree | 57f2b22d4b3bbaf21a9849f724fd5f8c86544c4e /doc | |
| parent | 837ff4ee24cb302da0a17e07ebab8ab1ef9204ca (diff) | |
Add OS2ECP benchmark.
Turns out decompressing a point is ~50x slower than checking (x,y)
is on the curve. Update relnote accordingly.
Diffstat (limited to 'doc')
| -rw-r--r-- | doc/news.rst | 8 |
1 files changed, 2 insertions, 6 deletions
diff --git a/doc/news.rst b/doc/news.rst index ad3015082..80a8457e3 100644 --- a/doc/news.rst +++ b/doc/news.rst @@ -13,12 +13,8 @@ Version 1.11.33, Not Yet Released * Add support for the TLS Supported Point Formats Extension from RFC 4492. Adds TLS::Policy::use_ecc_point_compression policy option. If supported on both - sides, ECC points can be sent in compressed format, which both saves a few - bytes on the wire and is an inexpensive way of avoiding invalid curve attacks. - For uncompressed points Botan already checks that the point is on the curve so - invalid curve attacks are not possible in either situation, but the point - decompression will typically be cheaper than verifying the point is on the - curve. (GH #645) + sides, ECC points can be sent in compressed format which saves a few bytes + during the handshake. (GH #645) * Fix entropy source selection bug on Windows, which caused the CryptoAPI entropy source to be not available under its normal name "win32_cryptoapi" but |