Async cert validation for 1.3

Async cert validation for 1.3

[ekr-rietveld, 2016-05-21 19:23:14]

LGTM. Yes, I think we should find out how to uplift this to Nightly. Maybe roll
3.24.1 with Tim's changes too?

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
File external_tests/ssl_gtest/libssl_internals.c (right):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
external_tests/ssl_gtest/libssl_internals.c:102: sslSocket *ss =
ssl_FindSocket(fd);
Check for error.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
File external_tests/ssl_gtest/libssl_internals.h (left):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
external_tests/ssl_gtest/libssl_internals.h:31: 
This is a weird comment.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
File external_tests/ssl_gtest/ssl_loopback_unittest.cc (left):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1182: }
Why this change?

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
File external_tests/ssl_gtest/ssl_loopback_unittest.cc (right):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1283: // messages.  The
Finished message is then processed with the sec.
sec.?

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1309: case 3:
This is pretty confusing. Packet #1 is ServerHello...Finished?, so #2 is
ServerHello...CertificateVerify and #3 is Finished?

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1352: return SECWouldBlock;
So in this case, the situation is that the whole server first flight has been
processed?

[mt, 2016-05-21 20:46:14]

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
File external_tests/ssl_gtest/libssl_internals.c (right):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/libs...
external_tests/ssl_gtest/libssl_internals.c:102: sslSocket *ss =
ssl_FindSocket(fd);
On 2016/05/21 19:23:14, ekr-webrtc wrote:
> Check for error.

Done.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
File external_tests/ssl_gtest/ssl_loopback_unittest.cc (left):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1182: }
On 2016/05/21 19:23:14, ekr-webrtc wrote:
> Why this change?

I originally thought that I would copy this to the new filter code that I added.

But then I realized that Handshake() works well enough if you make the tweak
that I made.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
File external_tests/ssl_gtest/ssl_loopback_unittest.cc (right):

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1283: // messages.  The
Finished message is then processed with the sec.
On 2016/05/21 19:23:14, ekr-webrtc wrote:
> sec.?

Fixed.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1309: case 3:
On 2016/05/21 19:23:14, ekr-webrtc wrote:
> This is pretty confusing. Packet #1 is ServerHello...Finished?, so #2 is
> ServerHello...CertificateVerify and #3 is Finished?

I'll add a few comments.

https://codereview.appspot.com/293470043/diff/1/external_tests/ssl_gtest/ssl_...
external_tests/ssl_gtest/ssl_loopback_unittest.cc:1352: return SECWouldBlock;
On 2016/05/21 19:23:14, ekr-webrtc wrote:
> So in this case, the situation is that the whole server first flight has been
> processed?

Yes.  More comments added.