Notes from Emilia (review of RT3001):
The other point in that ticket leads to something far more important: that test
is bust. It's setting the first parameter of RSA_private_decrypt - length of
the ciphertext - to |num|, however |num|, at that point, is plen, i.e., the
length of the plaintext. So that test does not exercise data corruption the way
it's meant to - it fails because the passed-in ciphertext has bad length.
(Truncation is a good test, too, but this test is confused.)

Also, if it's meant to test single-byte mutations, it isn't doing exactly that
either as it's not resetting the bytes to their original values.

So that needs fixing. Whether the test breaks early *on failure* is unimportant
imho. If it's too slow then we should cut some cases *always*, for example, not
test all 255 byte mutations.

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [email protected]
Automated List Manager                           [email protected]

Reply via email to