Fix hangs in Mandoline page cycler on Linux with --enable-multiprocess.

Fix hangs in Mandoline page cycler on Linux with --enable-multiprocess.

This was caused by content handlers sometimes dropping a request. The solution is to implement a simplified version of the two-phase shutdown used by apps. First, the ContentHandler implementation needs to keep a reference for the app's lifetime. Then every time StartApplication is called, a callback is passed along to be run when the content handler instance is destructed. When the ContentHandlerConnection notices that all the instances are deleted, it destructs itself. This releases the ContentHandler implementation which releases the app lifetime reference.

BUG=544153, 478251
Committed: https://crrev.com/fa71930f2532abee206f44755c2463b7aba9d5cb
Cr-Commit-Position: refs/heads/master@{#357301}

[jam, 2015-10-31 02:08:45]

jam@chromium.org changed reviewers:
+ yzshen@chromium.org

[yzshen1, 2015-10-31 22:55:08]

LGTM with 2 nits.

Thanks!

https://codereview.chromium.org/1431573002/diff/100001/mojo/application/publi...
File mojo/application/public/interfaces/content_handler.mojom (right):

https://codereview.chromium.org/1431573002/diff/100001/mojo/application/publi...
mojo/application/public/interfaces/content_handler.mojom:15: // binding, which
should cause the strongly-bound ContentHandler implementation
nit: binding -> interface pointer, please (for binding refers to the impl side
of the interface).

https://codereview.chromium.org/1431573002/diff/100001/mojo/fetcher/about_fet...
File mojo/fetcher/about_fetcher_unittest.cc (right):

https://codereview.chromium.org/1431573002/diff/100001/mojo/fetcher/about_fet...
mojo/fetcher/about_fetcher_unittest.cc:57: const mojo::Callback<void()>&
destruct_callback) override {
nit: this is inside the mojo namespace, so "mojo::" could be omitted. (here and
elsewhere)

[jam, 2015-11-02 02:50:12]

Patchset #7 (id:120001) has been deleted

[jam, 2015-11-02 02:50:18]

https://codereview.chromium.org/1431573002/diff/100001/mojo/application/publi...
File mojo/application/public/interfaces/content_handler.mojom (right):

https://codereview.chromium.org/1431573002/diff/100001/mojo/application/publi...
mojo/application/public/interfaces/content_handler.mojom:15: // binding, which
should cause the strongly-bound ContentHandler implementation
On 2015/10/31 22:55:08, yzshen1 wrote:
> nit: binding -> interface pointer, please (for binding refers to the impl side
> of the interface).

Done.

https://codereview.chromium.org/1431573002/diff/100001/mojo/fetcher/about_fet...
File mojo/fetcher/about_fetcher_unittest.cc (right):

https://codereview.chromium.org/1431573002/diff/100001/mojo/fetcher/about_fet...
mojo/fetcher/about_fetcher_unittest.cc:57: const mojo::Callback<void()>&
destruct_callback) override {
On 2015/10/31 22:55:08, yzshen1 wrote:
> nit: this is inside the mojo namespace, so "mojo::" could be omitted. (here
and
> elsewhere)

Done.

[jam, 2015-11-02 02:50:22]

The CQ bit was checked by jam@chromium.org

[jam, 2015-11-02 02:50:22]

The patchset sent to the CQ was uploaded after l-g-t-m from yzshen@chromium.org
Link to the patchset: https://codereview.chromium.org/1431573002/#ps140001
(title: "nits")

[commit-bot: I haz the power, 2015-11-02 02:50:37]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1431573002/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1431573002/140001

[commit-bot: I haz the power, 2015-11-02 03:45:40]

Committed patchset #7 (id:140001)

[commit-bot: I haz the power, 2015-11-02 03:46:41]

Patchset 7 (id:??) landed as
https://crrev.com/fa71930f2532abee206f44755c2463b7aba9d5cb
Cr-Commit-Position: refs/heads/master@{#357301}