QUIC - add instrumentation to HPACK.

QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=

Committed: https://crrev.com/ff8d65d99ceb8cb9e19c29281defdeffb0a389fa
Cr-Commit-Position: refs/heads/master@{#395160}

[Buck, 2016-05-18 19:47:39]

The CQ bit was checked by ckrasic@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-05-18 19:48:27]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1989003002/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1989003002/80001

[commit-bot: I haz the power, 2016-05-18 20:17:10]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-05-18 20:17:11]

Dry run: Try jobs failed on following builders:
  mac_chromium_gn_rel on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_gn_r...)
  win_chromium_rel_ng on tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[Buck, 2016-05-20 19:34:54]

Description was changed from

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging.

BUG=
==========

to

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging.

BUG=
==========

[Buck, 2016-05-20 19:34:54]

ckrasic@chromium.org changed reviewers:
+ rch@chromium.org

[Buck, 2016-05-20 19:36:17]

Description was changed from

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging.

BUG=
==========

to

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=
==========

[Buck, 2016-05-20 19:37:41]

This is a straight up merge of the internal change.   The chromium specific code
for UMA logging is in a followup (https://codereview.chromium.org/1987913002/)

[Ryan Hamilton, 2016-05-20 20:50:53]

The CQ bit was checked by rch@chromium.org

[Ryan Hamilton, 2016-05-20 20:50:53]

lgtm

Woo hoo! Facade design pattern for the win!

[commit-bot: I haz the power, 2016-05-20 20:51:16]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1989003002/200001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1989003002/200001

[commit-bot: I haz the power, 2016-05-20 20:57:04]

Description was changed from

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=
==========

to

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=
==========

[commit-bot: I haz the power, 2016-05-20 20:57:05]

Committed patchset #11 (id:200001)

[commit-bot: I haz the power, 2016-05-20 20:59:22]

Description was changed from

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=
==========

to

==========
QUIC - add instrumentation to HPACK.

Measure delta between time of creation and use for dynamic table entries.

This instrumentation is help sanity check a proposed scheme to eliminate
HPACK induced HOL blocking in QUIC.  The hypothesis is that most indexed
representations occur some non-negligible time after the entries are
added to the table in the first place.  This is somewhat implied by data
that shows 98% of QUIC sessions do not experience HOL blocking.  If data
from this new instrumentation shows the hypothesis false, then the
proposed scheme is not sound.  [ Note that the opposite inference is not
assumed. ]

This change introduces |HpackHeaderTable::DebugVisitorInterface|
that surfaces dynamic table events with |OnNewEntry()| and
|OnUseEntry()|, plumbed via |SpdyFramer| to
|QuicHeadersStream|. Generic timekeeping code isn't available
between chromium and the internal codebase, but QUIC code has
|QuicTime| which is available both codebases.  So the new
DebugVisitorInterface uses int64_t as opaque time type, which is
marshalled to/from |QuicTime| in |QuicHeadersStream|, the actual
timekeeping lives on the QUIC side, and uses existing |QuicTime|
and |QuicClock| facilities.  |QuicHeaderStream::HpackDebugVisitor|
wraps HHT:DVI providing a |OnUseEntry(QuicTime::Delta elapsed)|
time suitable for use in tests and the followup change that will
do UMA logging (https://codereview.chromium.org/1987913002/).

Merge internal change: 122853032

BUG=

Committed: https://crrev.com/ff8d65d99ceb8cb9e19c29281defdeffb0a389fa
Cr-Commit-Position: refs/heads/master@{#395160}
==========

[commit-bot: I haz the power, 2016-05-20 20:59:23]

Patchset 11 (id:??) landed as
https://crrev.com/ff8d65d99ceb8cb9e19c29281defdeffb0a389fa
Cr-Commit-Position: refs/heads/master@{#395160}