IndexedDB: Ensure that blobs in unused prefetches are acked

IndexedDB: Ensure that blobs in unused prefetches are acked

When the browser sends a blob reference to the renderer it holds an
extra reference until the renderer "acks" receipt, to keep the blob
alive. Normally this is done when a "result" containing blobs is
handed off to a blink::IDBRequest.

When we "prefetch" for a cursor, multiple results are sent from the
browser to the renderer at once, but only seen by blink::IDBRequest
when script actually requests them. If script never does (e.g. it
abandons a cursor midway through iteration) then the blobs would never
be acked.

Add blob acking to the cleanup step when prefetch entries are dropped.

BUG=457498
R=cmumford@chromium.org

Committed: https://crrev.com/0dd33709e0a2cbf5c30d76d1dd3367fce9afa6ed
Cr-Commit-Position: refs/heads/master@{#316688}

[jsbell, 2015-02-17 17:54:53]

cmumford@ - please take a look?

[cmumford, 2015-02-17 18:13:49]

https://codereview.chromium.org/924033003/diff/1/content/test/data/indexeddb/...
File content/test/data/indexeddb/blob_did_ack_prefetch.html (right):

https://codereview.chromium.org/924033003/diff/1/content/test/data/indexeddb/...
content/test/data/indexeddb/blob_did_ack_prefetch.html:14: var store_name =
'blobs_ack';
Are we guaranteed this name won't collide with the store in
content/test/data/indexeddb/blob_did_ack.html?

[jsbell, 2015-02-17 18:15:11]

https://codereview.chromium.org/924033003/diff/1/content/test/data/indexeddb/...
File content/test/data/indexeddb/blob_did_ack_prefetch.html (right):

https://codereview.chromium.org/924033003/diff/1/content/test/data/indexeddb/...
content/test/data/indexeddb/blob_did_ack_prefetch.html:14: var store_name =
'blobs_ack';
On 2015/02/17 18:13:49, cmumford wrote:
> Are we guaranteed this name won't collide with the store in
> content/test/data/indexeddb/blob_did_ack.html?

We should not be preserving state across tests, but I'll change it anyway.

[jsbell, 2015-02-17 18:23:38]

> We should not be preserving state across tests, but I'll change it anyway.

Also, the database name is created based on the name of the HTML file (see
common.js) so there's no chance of colliding anyway. But still - I tweaked the
store name.

[cmumford, 2015-02-17 18:24:11]

lgtm

[jsbell, 2015-02-17 18:24:33]

The CQ bit was checked by jsbell@chromium.org

[commit-bot: I haz the power, 2015-02-17 18:25:30]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/924033003/20001

[commit-bot: I haz the power, 2015-02-17 20:35:52]

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

[commit-bot: I haz the power, 2015-02-17 20:35:53]

Try jobs failed on following builders:
  android_rel_tests_recipe on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/android_rel_tes...)

[jsbell, 2015-02-17 21:02:06]

The CQ bit was checked by jsbell@chromium.org

[commit-bot: I haz the power, 2015-02-17 21:02:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/924033003/20001

[commit-bot: I haz the power, 2015-02-17 22:56:08]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-02-17 22:56:34]

Patchset 2 (id:??) landed as
https://crrev.com/0dd33709e0a2cbf5c30d76d1dd3367fce9afa6ed
Cr-Commit-Position: refs/heads/master@{#316688}