Update net/third_party/nss to NSS 3.13.3.

Update net/third_party/nss to NSS 3.13.3.

The following patches have been upstreamed:
- net/third_party/nss/patches/handshakeshortwrite.patch
- net/third_party/nss/patches/cbcrandomiv.patch
- net/third_party/nss/patches/nextproto.patch
- portions of patches/cachecerts.patch that add certificates to
  ss->ssl3.peerCertChain in the right order.
- portions of net/third_party/nss/patches/clientauth.patch that
  fix NSS bug 616757.

I omitted the net/third_party/nss/patches/cachedinfo.patch because
Chrome isn't using the TLS cached info extension and I wanted to
maintain fewer patches.  We can add it back later.

R=rsleevi@chromium.org,agl@chromium.org
BUG=116617
TEST=Unit tests should pass. Manual tests of SSL client auth and
origin-bound certs.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=124862

[Ryan Sleevi, 2012-03-02 01:30:18]

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/README.c...
File net/third_party/nss/README.chromium (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/README.c...
net/third_party/nss/README.chromium:3: 
Version: 3.13.3
Security Critical: Yes

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/README.c...
net/third_party/nss/README.chromium:56: patches/restartclientauth.patch
I thought bsmith upstreamed this already?

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssl3...
File net/third_party/nss/ssl/ssl3con.c (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssl3...
net/third_party/nss/ssl/ssl3con.c:5950: * application data until we have
authenticated the server's certificate;
Is this true, especially in the context of false start?

(Updated) Apparently line 8306 answers this - but that's a long way to read to
find that subtlety out.

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
File net/third_party/nss/ssl/sslimpl.h (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
net/third_party/nss/ssl/sslimpl.h:872: /* chain while we are trying to validate
it.   */
Does this mean |peerCertChain| now contains the leaf?

[wtc, 2012-03-02 23:06:17]

rsleevi: thank you for your comments.  Please review patch set 3.

agl: just FYI.  It should suffice to just look at the CL's
description and the following files:
  net/socket/ssl_client_socket_nss.h
  net/socket/ssl_client_socket_nss.cc
  net/third_party/nss/README.chromium
  net/third_party/nss/patches/applypatches.sh

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/README.c...
File net/third_party/nss/README.chromium (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/README.c...
net/third_party/nss/README.chromium:56: patches/restartclientauth.patch

On 2012/03/02 01:30:18, Ryan Sleevi wrote:
> I thought bsmith upstreamed this already?

No.  What bsmith implemented is the related function to
restart a handshake after verifying the server certificate.

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssl3...
File net/third_party/nss/ssl/ssl3con.c (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssl3...
net/third_party/nss/ssl/ssl3con.c:5950: * application data until we have
authenticated the server's certificate;

On 2012/03/02 01:30:18, Ryan Sleevi wrote:
> Is this true, especially in the context of false start?
> 
> (Updated) Apparently line 8306 answers this - but that's a long way to read to
> find that subtlety out.

Yes.  False start is disabled on line 8306 as a precaution.
We will need to come up with a better solution that allows
us to do false start safely.  I think that simply adding
    || ss->opt.enableFalseStart
to the conditional expression on line 5970 will accomplish
that.  Agreed?  (You don't need to check that in this
review.)

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
File net/third_party/nss/ssl/sslimpl.h (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
net/third_party/nss/ssl/sslimpl.h:872: /* chain while we are trying to validate
it.   */

On 2012/03/02 01:30:18, Ryan Sleevi wrote:
> Does this mean |peerCertChain| now contains the leaf?

Good question.  I checked the changes to ssl3_HandleCertificate
in NSS 3.13.2.  There is no change in this area.  I think
the comment bsmith removed is useful info, so I'll write a
patch to add it back.

http://codereview.chromium.org/9558017/diff/14001/net/socket/ssl_client_socke...
File net/socket/ssl_client_socket_nss.cc (left):

http://codereview.chromium.org/9558017/diff/14001/net/socket/ssl_client_socke...
net/socket/ssl_client_socket_nss.cc:940: #ifdef SSL_NEXT_PROTO_NEGOTIATED

In the NSS upstream, SSL_NEXT_PROTO_NEGOTIATED is defined as
an enum constant rather than a macro.  So we can't use this
ifdef, and there is no macro that we can use to detect the
NPN feature.

http://codereview.chromium.org/9558017/diff/14001/net/socket/ssl_client_socke...
File net/socket/ssl_client_socket_nss.cc (right):

http://codereview.chromium.org/9558017/diff/14001/net/socket/ssl_client_socke...
net/socket/ssl_client_socket_nss.cc:949: rv = SSL_OptionSet(nss_fd_,
SSL_CBC_RANDOM_IV, false_start_enabled);

The SSL_CBC_RANDOM_IV option (SSL 1/n-1 record splitting) is
enabled by default.  But we are linking it to False Start to
work around Brocade SSL terminators' intolerance to record
splitting (http://codereview.chromium.org/8137027).

Our workaround was in ssl3con.c because the
SSL_CBC_RANDOM_IV option wasn't available.  Now we can do
it here.

http://codereview.chromium.org/9558017/diff/14001/net/socket/ssl_client_socke...
net/socket/ssl_client_socket_nss.cc:1397: #ifdef SSL_ENABLE_CACHED_INFO

This ifdef is necessary because SSL_SetPredictedPeerCertificates
is defined in cachedinfo.patch, which I removed in this CL.

[agl, 2012-03-02 23:17:57]

I checked the requested files. LGTM.

[Ryan Sleevi, 2012-03-02 23:39:25]

lgtm

[wtc, 2012-03-03 02:34:12]

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
File net/third_party/nss/ssl/sslimpl.h (right):

http://codereview.chromium.org/9558017/diff/7001/net/third_party/nss/ssl/ssli...
net/third_party/nss/ssl/sslimpl.h:872: /* chain while we are trying to validate
it.   */

rsleevi: it turns out the deleted comment comes from the
cachedinfo.patch, which is why it's gone.  bsmith didn't
delete it.

[commit-bot: I haz the power, 2012-03-03 13:23:43]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/wtc@chromium.org/9558017/15052

[commit-bot: I haz the power, 2012-03-03 15:02:25]

Change committed as 124862