On Wed, 11 Aug 2021 21:20:42 +0200,
Kurt Roeckx wrote:
>
> There are a lot of things we accept in a certificate we shouldn't.
PR #16027 isn't about what our code accepts, but about what it
*produces*.
At the very least, I see an interop problem, because when certain
necessary structure values are missing, the DER (and PEM) encoding
will be invalid for the declared content type.
Just take the program from issue openssl/openssl#16026, but change the
i2d_RSAPrivateKey() call to a PEM_write_RSAPrivateKey() call. This is
a run again OpenSSL 1.1.1k (which is the version before this patch):
: ; ./foo
-----BEGIN RSA PRIVATE KEY-----
MA0CAQACAwHiQAIDAQAB
-----END RSA PRIVATE KEY-----
Failed 'PEM_write_RSAPrivateKey(stdout, rsa, NULL, NULL, 0, NULL, NULL) <=
0'
: levitte@lapdog:~/gitwrk/openssl.net/official/1.1.1
: ; ./foo | openssl asn1parse -i
Failed 'PEM_write_RSAPrivateKey(stdout, rsa, NULL, NULL, 0, NULL, NULL) <=
0'
0:d=0 hl=2 l= 13 cons: SEQUENCE
2:d=1 hl=2 l= 1 prim: INTEGER :00
5:d=1 hl=2 l= 3 prim: INTEGER :01E240
10:d=1 hl=2 l= 3 prim: INTEGER :010001
That's a badly formatted RSAPrivateKey (it looks like a RSAPublicKey).
Cheers,
Richard
--
Richard Levitte [email protected]
OpenSSL Project http://www.openssl.org/~levitte/
