Add max_bandwidth and max_bandwidth_timestamp to QUIC source address

Add max_bandwidth and max_bandwidth_timestamp to QUIC source address
token.

Merge internal change: 73055131

R=rch@chromium.org, wtc@chromium.org

[Ryan Hamilton, 2014-08-13 16:28:00]

lgtm

[wtc, 2014-08-14 01:49:10]

Patch set 1 LGTM. I believe the new code has the year 2038 bug. Please avoid it.

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
File net/quic/crypto/source_address_token.h (right):

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
net/quic/crypto/source_address_token.h:86: // The maximum bandwidth seen to the
client, not necessarily the latest.

Nit: seen to the client => seen by the client ?

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
net/quic/crypto/source_address_token.h:90: int32
max_bandwidth_timestamp_seconds_;

IMPORTANT: this signed 32-bit integer will overflow in year 2038. This is the
Unix year 2038 problem. Can this be an int64?

[ramant (doing other things), 2014-08-14 19:48:04]

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
File net/quic/crypto/source_address_token.h (right):

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
net/quic/crypto/source_address_token.h:86: // The maximum bandwidth seen to the
client, not necessarily the latest.
On 2014/08/14 01:49:10, wtc wrote:
> 
> Nit: seen to the client => seen by the client ?

Done.

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
net/quic/crypto/source_address_token.h:90: int32
max_bandwidth_timestamp_seconds_;
On 2014/08/14 01:49:10, wtc wrote:
> 
> IMPORTANT: this signed 32-bit integer will overflow in year 2038. This is the
> Unix year 2038 problem. Can this be an int64?

Filed b/17035772. Chromium is not yet using this code. Will fix in the internal
source also.

[ramant (doing other things), 2014-08-14 19:52:09]

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
File net/quic/crypto/source_address_token.h (right):

https://codereview.chromium.org/463093003/diff/1/net/quic/crypto/source_addre...
net/quic/crypto/source_address_token.h:90: int32
max_bandwidth_timestamp_seconds_;
On 2014/08/14 01:49:10, wtc wrote:
> 
> IMPORTANT: this signed 32-bit integer will overflow in year 2038. This is the
> Unix year 2038 problem. Can this be an int64?

Done.

[wtc, 2014-08-14 22:36:26]

Review comments on patch set 2:

https://codereview.chromium.org/463093003/diff/20001/net/quic/crypto/source_a...
File net/quic/crypto/source_address_token.h (right):

https://codereview.chromium.org/463093003/diff/20001/net/quic/crypto/source_a...
net/quic/crypto/source_address_token.h:90: int64
max_bandwidth_timestamp_seconds_;

The getter and setter methods also need to use the int64 type. Also, it may be
better to use one of the types defined in quic_time.h (or quic_clock?) instead.

[ramant (doing other things), 2014-08-14 23:40:45]

Thanks very much Want-Teh. Updated the "Landing Recent QUIC changes" to have
this change.

https://codereview.chromium.org/463093003/diff/20001/net/quic/crypto/source_a...
File net/quic/crypto/source_address_token.h (right):

https://codereview.chromium.org/463093003/diff/20001/net/quic/crypto/source_a...
net/quic/crypto/source_address_token.h:90: int64
max_bandwidth_timestamp_seconds_;
On 2014/08/14 22:36:26, wtc wrote:
> 
> The getter and setter methods also need to use the int64 type. Also, it may be
> better to use one of the types defined in quic_time.h (or quic_clock?)
instead.

Changed get/set methods to use int64. 

Would like to use protobuf (ala internal source tree) and delete this file. Will
bring it up if we could use QuicWallTime in the internal source tree.