Normalize hostnames before searching for HSTS/HPKP preloads

Normalize hostnames before searching for HSTS/HPKP preloads

The HSTS/HPKP preload set is pre-normalized at construction time.
Since the queries come from the GURL, not from the DNS layer,
we need to normalize the hostname before scanning for preloads.
This used to be handled by CanonicalizeHost(), which used the
same mechanism as the resolver, but the storage of the preloads
has changed to be more efficient, and thus no longer uses the
resolver-normalized form.

BUG=461481
R=davidben@chromium.org

Committed: https://crrev.com/c2dd599bdd35753da237172dc37a447e9d65c463
Cr-Commit-Position: refs/heads/master@{#330834}

[Ryan Sleevi, 2015-05-20 18:31:54]

For your review

[davidben, 2015-05-20 19:21:09]

Just a bunch of nits and the CanonicalizeHost thing.

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
File net/http/transport_security_state.cc (left):

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
net/http/transport_security_state.cc:301: 
Stray change?

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
File net/http/transport_security_state.cc (right):

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
net/http/transport_security_state.cc:508: // Normalize any trailing '.' used for
DNS suffix searches
Nit: period

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
net/http/transport_security_state.cc:519: // lower case
Nit: period

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
net/http/transport_security_state.cc:742: return !CanonicalizeHost(host).empty()
&& DecodeHSTSPreload(host, &result) &&
Would it be better to move the CanonicalizeHost check inside DecodeHSTSPreload?
Otherwise probably some new caller will forget to do it. And, as I understand,
that's closer to the old behavior?

[Ryan Sleevi, 2015-05-20 19:26:40]

Done. I split this into two patchsets - one in which I just make
CanonicalizeHost a free function in //net, which required pre-declaring it, and
the second in which I move this (and the related, supporting free functions)
into the unnamed namespace to avoid any ODR issues.

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
File net/http/transport_security_state.cc (left):

https://codereview.chromium.org/1149753002/diff/1/net/http/transport_security...
net/http/transport_security_state.cc:301: 
On 2015/05/20 19:21:09, David Benjamin wrote:
> Stray change?

Readability ;)

[davidben, 2015-05-20 19:39:13]

lgtm

[Ryan Sleevi, 2015-05-20 19:42:43]

The CQ bit was checked by rsleevi@chromium.org

[commit-bot: I haz the power, 2015-05-20 19:44:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1149753002/40001

[commit-bot: I haz the power, 2015-05-20 23:29:14]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-05-20 23:30:50]

Patchset 3 (id:??) landed as
https://crrev.com/c2dd599bdd35753da237172dc37a447e9d65c463
Cr-Commit-Position: refs/heads/master@{#330834}