Relax DCHECKs regarding Threads and RefCounts in iOS' HttpServer.

Relax DCHECKs regarding Threads and RefCounts in iOS' HttpServer.

BUG=707764

[jif, 2017-04-03 14:05:34]

Description was changed from

==========
Relax DCHECKs regarding Threads and RefCounts in iOS' HttpServer.

BUG=707764
==========

to

==========
Relax DCHECKs regarding Threads and RefCounts in iOS' HttpServer.

BUG=707764
==========

[jif, 2017-04-03 14:05:34]

jif@chromium.org changed reviewers:
+ sdefresne@chromium.org

[jif, 2017-04-03 14:05:46]

ptal

[sdefresne, 2017-04-03 14:18:16]

lgtm to unblock the tree

The suggested changes should be done in a separate CL probably.

https://codereview.chromium.org/2794933002/diff/1/ios/web/public/test/http_se...
File ios/web/public/test/http_server.mm (right):

https://codereview.chromium.org/2794933002/diff/1/ios/web/public/test/http_se...
ios/web/public/test/http_server.mm:84: // TODO(crbug.com/707764): Verify that
the relaxation of the cross-thread
I think that a correct fix would be to:

1. split GetResponseProviderForRequest in two
  - GetResponseProviderForRequestImpl that does all the work except the locking,
  - GetResponseProviderForRequest that take the lock then call
GetResponseProviderForRequestImpl
2. changing here to acquire the lock, then call
GetResponseProviderForRequestImpl
3. changing the comment to the following

  // Relax the cross-thread access restriction to non-thread-safe RefCount.
  // |provider_list_lock_| protects the access to the
  // RefCountedResponseProviderWrapper instances.

https://codereview.chromium.org/2794933002/diff/1/ios/web/public/test/http_se...
ios/web/public/test/http_server.mm:234: // TODO(crbug.com/707764): Verify that
the relaxation of the cross-thread
I think you can change this to:

  // Relax the cross-thread access restriction to non-thread-safe RefCount.
  // |provider_list_lock_| protects the access to the
  // RefCountedResponseProviderWrapper instances.

[baxley, 2017-04-03 14:21:08]

baxley@chromium.org changed reviewers:
+ baxley@chromium.org

[baxley, 2017-04-03 14:21:09]

+haungml@,michaeldo@ who did similar changes on Friday.

[jif, 2017-04-03 14:30:11]

On 2017/04/03 14:21:09, baxley wrote:
> +haungml@,michaeldo@ who did similar changes on Friday.

ooops, michaeldo sent me his CL friday but I forgot about it since then :(

[jif, 2017-04-03 14:34:12]

On 2017/04/03 14:30:11, jif wrote:
> On 2017/04/03 14:21:09, baxley wrote:
> > +haungml@,michaeldo@ who did similar changes on Friday.
> 
> ooops, michaeldo sent me his CL friday but I forgot about it since then :(

going to reland his CL ( https://codereview.chromium.org/2783173004 ) instead.
Will put sdefresne's suggestions in the bug.

[jif, 2017-04-03 14:39:08]

The CQ bit was checked by jif@chromium.org

[jif, 2017-04-03 14:39:09]

The patchset sent to the CQ was uploaded after l-g-t-m from
sdefresne@chromium.org
Link to the patchset: https://codereview.chromium.org/2794933002/#ps20001
(title: "Fix broken ios_chrome_web_egtests.")

[commit-bot: I haz the power, 2017-04-03 14:39:23]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-03 14:51:51]

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

[commit-bot: I haz the power, 2017-04-03 14:51:52]

Failed to apply the patch.
On branch working_branch
Your branch is up-to-date with 'origin/refs/heads/master'.
nothing to commit, working tree clean

[Eugene But (OOO till 7-30), 2017-04-03 17:03:36]

eugenebut@chromium.org changed reviewers:
+ eugenebut@chromium.org

[Eugene But (OOO till 7-30), 2017-04-03 17:03:36]

Just FYI: Menglu is working of dropping GCDWebServer support, so we may not even
need a sophisticated fix for threading problem.