Reduce copy on QuotaFileIO::Write

Reduce copy on QuotaFileIO::Write.

FileUtilProxy::Write makes a copy of the data before posting the write job, and it is one of the bottlenecks of the write operation on pepper.
Though, we can safely omit the copy in QuotaFileIO::Write.

BUG=168744
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=177125

[kinuko, 2013-01-04 10:28:59]

lgtm, nice.

(Please update BUG= line before asking owner's review)

[tzik, 2013-01-08 04:48:54]

Thanks, Kinuko-san.

Brett, could you take a look to this?

[brettw, 2013-01-09 20:17:10]

https://codereview.chromium.org/11615036/diff/1/webkit/plugins/ppapi/quota_fi...
File webkit/plugins/ppapi/quota_file_io.cc (right):

https://codereview.chromium.org/11615036/diff/1/webkit/plugins/ppapi/quota_fi...
webkit/plugins/ppapi/quota_file_io.cc:131: base::PlatformFileError error =
bytes_written >= 0 ?
What happens if you write 0 bytes? It looks like before this may have succeeded,
and now it will report failure.

[tzik, 2013-01-10 04:02:36]

Thanks, updated.

https://codereview.chromium.org/11615036/diff/1/webkit/plugins/ppapi/quota_fi...
File webkit/plugins/ppapi/quota_file_io.cc (right):

https://codereview.chromium.org/11615036/diff/1/webkit/plugins/ppapi/quota_fi...
webkit/plugins/ppapi/quota_file_io.cc:131: base::PlatformFileError error =
bytes_written >= 0 ?
On 2013/01/09 20:17:10, brettw wrote:
> What happens if you write 0 bytes? It looks like before this may have
succeeded,
> and now it will report failure.

Right, this changes its behavior for 0-byte write.
Fixed.

[brettw, 2013-01-14 06:02:24]

Just to double-check, what happens if a plugin says Write(<some buffer>, 0);
(zero byte write). What return error would they have gotten through the API?
This patch doesn't change that, right?

[tzik, 2013-01-14 12:42:16]

On 2013/01/14 06:02:24, brettw wrote:
> Just to double-check, what happens if a plugin says Write(<some buffer>, 0);
> (zero byte write). What return error would they have gotten through the API?
> This patch doesn't change that, right?

bytes_to_write == 0 case is handled at the first line of QuotaFileIO::Write, and
the API returns PP_ERROR_FAILED.
And, right, This patch doesn't change that part. But, this is not consistent
with non-quota managed case.
Should I fix in this patch?

[tzik, 2013-01-15 05:38:27]

On 2013/01/14 12:42:16, tzik wrote:
> On 2013/01/14 06:02:24, brettw wrote:
> > Just to double-check, what happens if a plugin says Write(<some buffer>, 0);
> > (zero byte write). What return error would they have gotten through the API?
> > This patch doesn't change that, right?
> 
> bytes_to_write == 0 case is handled at the first line of QuotaFileIO::Write,
and
> the API returns PP_ERROR_FAILED.
> And, right, This patch doesn't change that part. But, this is not consistent
> with non-quota managed case.
> Should I fix in this patch?

Ah sorry, I misunderstood.
Zero byte write is also handled in FileUtilProxy::Write and it is also an error
for it.
So, it behaves same to non-quota managed case.

[brettw, 2013-01-16 05:38:31]

LGTM

[tzik, 2013-01-16 07:05:07]

On 2013/01/16 05:38:31, brettw wrote:
> LGTM

Thanks!

[commit-bot: I haz the power, 2013-01-16 07:05:21]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/tzik@chromium.org/11615036/17002

[commit-bot: I haz the power, 2013-01-16 10:36:56]

Change committed as 177125