Implements keygen handler for openssl, but without private key persistence

Implements keygen handler for openssl, but without private key persistence


BUG=64917
TEST=net_unittest Keygen*

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=68470

[joth, 2010-12-02 17:19:42]

I'm still trying to get my head around the required storage model for private
keys (across RSAPrivateKey::FindFromPublicKeyInfo, KeygenHandler, CertDatabase,
and the ssl client socket itself) and how best to map this onto openssl (or
not), but this gets the tests passing in the meantime.
(Note I'll leave the bug open until keygen is supported in a more meaningful
way, i.e. with private key persistence of some sort)

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_unit...
File net/base/keygen_handler_unittest.cc (right):

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_unit...
net/base/keygen_handler_unittest.cc:21: #endif
I reordered these as it 2 out of 3 of the lint warning, and a recent email
thread on chromium-dev suggests this is now the preferred style.

[wtc, 2010-12-03 19:15:44]

Just a drive-by review comment.

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_open...
File net/base/keygen_handler_openssl.cc (right):

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:33: NETSCAPE_SPKI_sign(spki.get(), pkey,
EVP_md5());
Please add a comment to note that you're using MD5 signature here because
it's specified in HTML5, even though the SPKI structure's format allows you to
use SHA-1 signature.

[joth, 2010-12-06 18:04:20]

Thanks!

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_open...
File net/base/keygen_handler_openssl.cc (right):

http://codereview.chromium.org/5541002/diff/2001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:33: NETSCAPE_SPKI_sign(spki.get(), pkey,
EVP_md5());
On 2010/12/03 19:15:44, wtc wrote:
> Please add a comment to note that you're using MD5 signature here because
> it's specified in HTML5, even though the SPKI structure's format allows you to
> use SHA-1 signature.

Done.

[bulach, 2010-12-07 10:59:53]

LGTM

nits below:

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
File net/base/keygen_handler_openssl.cc (right):

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:20: DCHECK(key != NULL);
I think we don't normally DCHECK if we're de-referencing right afterwards..

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:24: // TODO(joth): Add an abstraction for
persisting OpenSSL private keys.
you may want to update http://crbug.com/64917 and link here

[joth, 2010-12-07 12:11:42]

Thanks, landing it now!

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
File net/base/keygen_handler_openssl.cc (right):

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:20: DCHECK(key != NULL);
On 2010/12/07 10:59:53, bulach wrote:
> I think we don't normally DCHECK if we're de-referencing right afterwards..

Done.

http://codereview.chromium.org/5541002/diff/8001/net/base/keygen_handler_open...
net/base/keygen_handler_openssl.cc:24: // TODO(joth): Add an abstraction for
persisting OpenSSL private keys.
On 2010/12/07 10:59:53, bulach wrote:
> you may want to update http://crbug.com/64917 and link here

Done.