Land Recent QUIC Changes.

Land Recent QUIC Changes.

Change QUIC's pacer to allow a 10 packet burst when the connection comes
out of quiescence.

This is in line with the kernel's qdisc implementation.

Merge internal change: 70357214
https://codereview.chromium.org/367853003/


QUIC varz to track measured and predicted bandwidth and rtt histograms
for video responses for which internal streamer provided bandwidth/rtt
estimates and QUIC connection has a reliable bandwidth measurement.

bandwidth/rtt varz histograms for video responses served over QUIC.
Not flag protected.

Merge internal change: 70327906
https://codereview.chromium.org/365833002/


Add comments describing changes introduced in each QUIC version.

Merge internal change: 70319561
https://codereview.chromium.org/369493002/


QUIC_VERSION_21: headers and crypto streams are now flow controlled at
the stream level, but do not contribute to connection level flow
control.

Merge internal change: 70318947
https://codereview.chromium.org/368803003/


Added following helper methods to track successful QUIC client port migrations:

+ peer_port_changed()
+ received_packet_manager()
+ sequence_number_of_last_sent_packet()
+ largest_observed()

The following is the original internal CL description:

Add two varz to track successful QUIC client port migrations:

  peer_port_changes
    -- Number of times we have seen the peer's port change
       mid-connection.

  successful_peer_port_migrations
    -- Successful connection migrations to a new peer port.

Add varz to track successful QUIC client port migrations. Protected
behind existing, enabled, FLAGS_..._quic_allow_port_migration.

Merge internal change: 70318257
https://codereview.chromium.org/366843002/


Test only change to improve QUIC's SendAlgorithmSimulator to support
multiple transfers with different types of send algorithms at once.

Intended to test BBR vs BBR and BBR vs TCP Cubic.

Merge internal change: 70254014

A fix to SendAlgorithmSimulator, which was making BbrSenderTest flaky.

Merge internal change: 70269655
https://codereview.chromium.org/367843002/


QUIC - added helper method config() to QuicClient.

The following is the description from the internal change:

	Fixing a bug in the (as yet unused UDP backend picking code) where we
	weren't setting the configured timeout correctly.

	Fixing a bug in the (flag-disabled) UDP backend picking code.

Merge internal change: 70239548
https://codereview.chromium.org/362873002/

R=rch@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=281079

[ramant (doing other things), 2014-07-02 03:10:51]

Hi Ryan,
  Stopped until recent strike register changes which requires merging of earlier
changes to strike register.

  Will continue merging all those CLs in another CL.

thanks
raman

[Ryan Hamilton, 2014-07-02 19:24:59]

lgtm

[ramant (doing other things), 2014-07-02 19:28:32]

The CQ bit was checked by rtenneti@chromium.org

[commit-bot: I haz the power, 2014-07-02 19:30:14]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/rtenneti@chromium.org/366863002/1

[commit-bot: I haz the power, 2014-07-02 19:47:03]

Change committed as 281079

[wtc, 2014-07-02 22:56:48]

Patch set 1 LGTM.

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h
File net/quic/quic_connection.h (right):

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h#n...
net/quic/quic_connection.h:309: virtual bool ProcessValidatedPacket();

Why does this method need to be virtual? I don't see any subclass override this
method.

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h#n...
net/quic/quic_connection.h:517: const QuicReceivedPacketManager&
received_packet_manager() {

Mark this method const?

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_protocol.h
File net/quic/quic_protocol.h (right):

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_protocol.h#new...
net/quic/quic_protocol.h:285: QUIC_VERSION_20 = 20,  // Independent
stream/session flow control windows.

In these two comments, did you mean "session level" or "connection level" flow
control?

[ramant (doing other things), 2014-07-16 19:43:36]

Fixed the comments in CL: https://codereview.chromium.org/394053003/ and will
merge that back into the internal source tree.

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h
File net/quic/quic_connection.h (right):

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h#n...
net/quic/quic_connection.h:309: virtual bool ProcessValidatedPacket();
On 2014/07/02 22:56:48, wtc wrote:
> 
> Why does this method need to be virtual? I don't see any subclass override
this
> method.

Our internal code override's this method. It is not overwritten in chrome.
Diverged chrome version of this file from internal source tree by removing the
virtual.

Done.

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_connection.h#n...
net/quic/quic_connection.h:517: const QuicReceivedPacketManager&
received_packet_manager() {
On 2014/07/02 22:56:48, wtc wrote:
> 
> Mark this method const?

Done.

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_protocol.h
File net/quic/quic_protocol.h (right):

https://codereview.chromium.org/366863002/diff/1/net/quic/quic_protocol.h#new...
net/quic/quic_protocol.h:285: QUIC_VERSION_20 = 20,  // Independent
stream/session flow control windows.
On 2014/07/02 22:56:48, wtc wrote:
> 
> In these two comments, did you mean "session level" or "connection level" flow
> control?

Done.