IndexedDB: Fix cursor prefetching edge cases

IndexedDB: Fix cursor prefetching edge cases

Cursor prefetch caches must be discarded when other
requests are made to ensure proper request sequencing.
Two edge cases were handled improperly if new records
was written just ahead of the cursor.

* A reset occurring before the prefetch results were
received would be ignored; since the newly records
weren't in the prefetch data, the cursor wouldn't see
them.

* A reset occurring after the results are received
would back up the cursor to before the new records,
even though the prefetch itself is a "continue"
and advanced past them already.

The fix is to reset the cache on receipt if necessary,
and to ensure the reset state accounts for the implicit
advance.

BUG=331570
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=243344

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=243421

[jsbell, 2014-01-03 23:04:00]

FYI - needs unit tests in addition the layout test can land blink-side. (Similar
to what's in the bug.)

[jsbell, 2014-01-03 23:05:11]

https://codereview.chromium.org/124323002/diff/1/content/browser/indexed_db/i...
File content/browser/indexed_db/indexed_db_cursor.cc (right):

https://codereview.chromium.org/124323002/diff/1/content/browser/indexed_db/i...
content/browser/indexed_db/indexed_db_cursor.cc:162: transaction_->ScheduleTask(
I haven't convinced myself one way or another whether resets should be tasks or
not.

[jsbell, 2014-01-06 17:51:55]

https://codereview.chromium.org/124323002/diff/1/content/browser/indexed_db/i...
File content/browser/indexed_db/indexed_db_cursor.cc (right):

https://codereview.chromium.org/124323002/diff/1/content/browser/indexed_db/i...
content/browser/indexed_db/indexed_db_cursor.cc:162: transaction_->ScheduleTask(
On 2014/01/03 23:05:11, jsbell wrote:
> I haven't convinced myself one way or another whether resets should be tasks
or
> not.

After further reflection, this shouldn't be necessary. The prefetch mechanism is
predicated on there not being any other requests outstanding while the prefetch
cache is being used. So resetting the cache synchronously is fine.

[jsbell, 2014-01-06 18:48:26]

dgrogan@ - please take a look?

Layout tests which more clearly demonstrate the problem are at:

https://codereview.chromium.org/105213009

[dgrogan, 2014-01-06 22:43:50]

lgtm

[commit-bot: I haz the power, 2014-01-06 22:55:52]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jsbell@chromium.org/124323002/140001

[commit-bot: I haz the power, 2014-01-07 12:00:31]

Retried try job too often on win_rel for step(s) browser_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2014-01-07 17:29:45]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jsbell@chromium.org/124323002/140001

[commit-bot: I haz the power, 2014-01-07 18:52:49]

Change committed as 243344

[jsbell, 2014-01-07 20:25:23]

Reverted due to a test-only leak. Fixed, resubmitting.

[commit-bot: I haz the power, 2014-01-07 20:25:57]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jsbell@chromium.org/124323002/370001

[commit-bot: I haz the power, 2014-01-07 22:55:07]

Change committed as 243421