Initialize the memory allocated by Vector<IsPod<T>>

Initialize the memory allocated by Vector<IsPod<T>>

For performance reasons
(see https://bugs.webkit.org/show_bug.cgi?id=101649#c5),
Vector used not to initialize the memory allocated for POD
types. This yielded to several use of uninitialized memory
in our code and potentially more of them so it makes sense
to just zero the allocated the memory.

This change just remove the previous initialization so
Vector now initialize the memory to 0 for POD.

BUG=364675, 158510
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=174384

[Julien - ping for review, 2014-05-19 16:27:43]

Attached is a general fix for a table issue (deeper context is
https://bugs.webkit.org/show_bug.cgi?id=101649). I haven't been able to get some
performance numbers from the tryjobs. Happy to do some local measurements if you
think it's needed to land the change.

[eseidel, 2014-05-20 00:48:11]

Erik Corry or mikhail are your best reviewers.

[Erik Corry, 2014-05-20 08:20:57]

LGTM, I assume you will keep an eye on the perf bots.

[Mikhail, 2014-05-20 11:20:36]

lgtm
let's see how much the performance will be affected with it. We might consider
putting back the "needsInitialization" flag and passing traits explicitly (as a
template argument) if needed (in order to avoid unnecessary initialization in
the performance-sensitive places).

[Erik Corry, 2014-05-20 12:08:15]

With Oilpan the memory is pre-zapped by the GC, so we will probably
want to avoid zeroing it again.

On Tue, May 20, 2014 at 1:20 PM,  <mikhail.pozdnyakov@intel.com> wrote:
> lgtm
> let's see how much the performance will be affected with it. We might
> consider
> putting back the "needsInitialization" flag and passing traits explicitly
> (as a
> template argument) if needed (in order to avoid unnecessary initialization
> in
> the performance-sensitive places).
>
>
> https://codereview.chromium.org/291933002/
>
> To unsubscribe from this group and stop receiving emails from it, send an
> email to blink-reviews+unsubscribe@chromium.org.



-- 
Erik Corry, Software Engineer
Google Denmark ApS - Frederiksborggade 20B, 1 sal,
1360 København K - Denmark - CVR nr. 28 86 69 84

To unsubscribe from this group and stop receiving emails from it, send an email
to blink-reviews+unsubscribe@chromium.org.

[Julien - ping for review, 2014-05-20 13:45:26]

On 2014/05/20 11:20:36, mikhail.pozdnyakov wrote:
> lgtm
> let's see how much the performance will be affected with it. We might consider
> putting back the "needsInitialization" flag and passing traits explicitly (as
a
> template argument) if needed (in order to avoid unnecessary initialization in
> the performance-sensitive places).

If that's too much of a performance penalty, I have a narrower fix for the
RenderTable issue. Reviewers in WebKit rightly pointed out that this was
peppering over the bigger issue, thus this patch.

[Julien - ping for review, 2014-05-20 13:45:37]

The CQ bit was checked by jchaffraix@chromium.org

[commit-bot: I haz the power, 2014-05-20 13:45:51]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/jchaffraix@chromium.org/291933002/20001

[commit-bot: I haz the power, 2014-05-20 15:04:37]

Change committed as 174384