[ServiceWorkerCache] Serialize ServiceWorkerCacheStorage operations

[ServiceWorkerCache] Serialize ServiceWorkerCacheStorage operations

There are correctness issues in the ServiceWorkerCacheStorage when
multiple operations occur on the same key in parallel. The various
steps of the operations can interleave. This CL serializes all storage
operations by placing all operations in a queue and calling the next
operation from the callback of the previous operation.

While this provides correctness, it is sub-optimal. We really only
need to serialize operations that affect the same key. A smarter
scheduler can be created in the future to implement that (see
https://crbug.com/451174).

What this CL does:

* Creates a new ServiceWorkerCacheScheduler class used by both
ServiceWorkerCache and ServiceWorkerCacheStorage
* Refactor ServiceWorkerCache to use the new scheduler
* Make ServiceWorkerCacheStorage run sequentially, using
  the scheduler

BUG=450697
Committed: https://crrev.com/d1292e44b91faf5098719af5e75dc04ec05e3a5a
Cr-Commit-Position: refs/heads/master@{#315022}

[jkarlin, 2015-02-02 17:55:35]

jkarlin@chromium.org changed reviewers:
+ michaeln@chromium.org

[jkarlin, 2015-02-02 17:55:36]

PTAL at everything, thanks!

[michaeln, 2015-02-03 19:14:22]

lgtm, makes sense to have Open() followed by Delete() result in no errors and no
cache

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
File content/browser/service_worker/service_worker_cache_storage.cc (right):

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:35: void
NoopClosure() {
use base::DoNothing() instead

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:424: void
ServiceWorkerCacheStorage::OpenCacheImpl(
probably should move this <<>>Impl method down in the .cc file so the order here
matches the order in the .h file

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:550:
scheduler_->ScheduleOperation(base::Bind(&NoopClosure));
you can use base::DoNothing() for this, see bind_helpers.h

[jkarlin, 2015-02-06 12:52:48]

Thanks!

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
File content/browser/service_worker/service_worker_cache_storage.cc (right):

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:35: void
NoopClosure() {
On 2015/02/03 19:14:22, michaeln wrote:
> use base::DoNothing() instead

Done.

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:424: void
ServiceWorkerCacheStorage::OpenCacheImpl(
On 2015/02/03 19:14:22, michaeln wrote:
> probably should move this <<>>Impl method down in the .cc file so the order
here
> matches the order in the .h file

Done.

https://codereview.chromium.org/867903005/diff/40001/content/browser/service_...
content/browser/service_worker/service_worker_cache_storage.cc:550:
scheduler_->ScheduleOperation(base::Bind(&NoopClosure));
On 2015/02/03 19:14:22, michaeln wrote:
> you can use base::DoNothing() for this, see bind_helpers.h

Thanks!

[jkarlin, 2015-02-06 12:59:38]

The CQ bit was checked by jkarlin@chromium.org

[commit-bot: I haz the power, 2015-02-06 12:59:55]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/867903005/60001

[commit-bot: I haz the power, 2015-02-06 13:55:13]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-02-06 13:56:15]

Patchset 4 (id:??) landed as
https://crrev.com/d1292e44b91faf5098719af5e75dc04ec05e3a5a
Cr-Commit-Position: refs/heads/master@{#315022}