IndexedDB: Report QuotaExceededError when commit fails due to disk full

IndexedDB: Report QuotaExceededError when commit fails due to disk full

If a transaction commit fails, check to see if the underlying cause is
"disk full" and report a more specific error. Usually the quota system
prevents us from even trying to commit to disk if space is tight, but
the disk could be filled by another process which the quota system is
blissfully ignorant of.

BUG=329628
R=cmumford@chromium.org

Committed: https://crrev.com/6f21843a3f5b773b5af93e8e1b92570e6f668779
Cr-Commit-Position: refs/heads/master@{#317403}

[jsbell, 2015-02-19 19:05:44]

cmumford@ - can you take an initial look?

I'm not ready to land this, but wanted to get your initial impressions on the
approach.

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
File content/browser/indexed_db/indexed_db_browsertest.cc (right):

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
content/browser/indexed_db/indexed_db_browsertest.cc:445: const int instance_num
= 7;
I want to document why this is 7. The test preamble does several commits during
the open phase (creating the database, committing the upgrade, ...). But 7 seems
crazy high.

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
File content/browser/indexed_db/indexed_db_database_error.h (right):

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
content/browser/indexed_db/indexed_db_database_error.h:17:
IndexedDBDatabaseError() {}
This is to allow:

IndexedDBDatabaseError error;
if (...)
  error = IndexedDBDatabaseError(...);
else
  error = IndexedDBDatabaseError(...);
callbacks->DoIt(error);

... which seemed to be the most readable, but I'm open to alternatives.

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/m...
File content/browser/indexed_db/mock_browsertest_indexed_db_class_factory.cc
(right):

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/m...
content/browser/indexed_db/mock_browsertest_indexed_db_class_factory.cc:73: //
TODO(jsbell): Consider parameterizing the failure mode.
Maybe as a follow-up?

[jsbell, 2015-02-19 21:26:26]

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
File content/browser/indexed_db/indexed_db_browsertest.cc (right):

https://codereview.chromium.org/932143003/diff/1/content/browser/indexed_db/i...
content/browser/indexed_db/indexed_db_browsertest.cc:445: const int instance_num
= 7;
On 2015/02/19 19:05:44, jsbell wrote:
> I want to document why this is 7. The test preamble does several commits
during
> the open phase (creating the database, committing the upgrade, ...). But 7
seems
> crazy high.

Done in latest PS. Also filed crbug.com/460236 to try and cut down the number.

[cmumford, 2015-02-20 17:50:17]

lgtm

https://codereview.chromium.org/932143003/diff/20001/content/browser/indexed_...
File content/browser/indexed_db/indexed_db_database_error.h (right):

https://codereview.chromium.org/932143003/diff/20001/content/browser/indexed_...
content/browser/indexed_db/indexed_db_database_error.h:17:
IndexedDBDatabaseError() {}
What do you think about getting in the habit of only declaring this here and
doing the:

IndexedDBDatabaseError::IndexedDBDatabaseError() = default;

down in the *.cc file?

https://codereview.chromium.org/932143003/diff/20001/content/browser/indexed_...
content/browser/indexed_db/indexed_db_database_error.h:28: uint16 code_ = 0;
yeah! I forgot this is now allowed.

[jsbell, 2015-02-20 18:25:32]

New patchsets have been uploaded after l-g-t-m from cmumford@chromium.org

[jsbell, 2015-02-20 18:27:02]

cmumford@ - one more look?

https://codereview.chromium.org/932143003/diff/20001/content/browser/indexed_...
File content/browser/indexed_db/indexed_db_database_error.h (right):

https://codereview.chromium.org/932143003/diff/20001/content/browser/indexed_...
content/browser/indexed_db/indexed_db_database_error.h:17:
IndexedDBDatabaseError() {}
On 2015/02/20 17:50:17, cmumford wrote:
> What do you think about getting in the habit of only declaring this here and
> doing the:
> 
> IndexedDBDatabaseError::IndexedDBDatabaseError() = default;
> 
> down in the *.cc file?

Done. Moved all the impls to (a new) .cc file.

[cmumford, 2015-02-20 20:55:49]

RSLGTM

[jsbell, 2015-02-20 21:07:10]

The CQ bit was checked by jsbell@chromium.org

[commit-bot: I haz the power, 2015-02-20 21:08:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/932143003/40001

[commit-bot: I haz the power, 2015-02-20 21:18:00]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-02-20 21:19:15]

Patchset 3 (id:??) landed as
https://crrev.com/6f21843a3f5b773b5af93e8e1b92570e6f668779
Cr-Commit-Position: refs/heads/master@{#317403}