Minor perf and size optimization - don't do asserts in release build

Minor perf and size optimization - don't do asserts in release build

Avoiding the pthread_getspecific call on every pointer access seems worth it (and it reduces templated code size)

BUG=None
TEST=None

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=108678

[willchan no longer on Chromium, 2011-11-04 14:39:30]

LGTM

On Fri, Nov 4, 2011 at 3:36 AM, <joth@chromium.org> wrote:

> Reviewers: willchan,
>
> Description:
> Minor perf and size optimization - don't do asserts in release build
>
> Avoiding the pthread_getspecific call on every pointer access seems worth
> it
> (and it reduces templated code size)
>
> BUG=None
> TEST=None
>
>
> Please review this at
http://codereview.chromium.**org/8465010/<http://codereview.chromium.org/8465...
>
> SVN Base:
svn://svn.chromium.org/chrome/**trunk/src<http://svn.chromium.org/chrome/trunk/src>
>
> Affected files:
>  M base/lazy_instance.h
>  M base/memory/singleton.h
>
>
> Index: base/lazy_instance.h
> diff --git a/base/lazy_instance.h b/base/lazy_instance.h
> index 496fc1f0404948986de277bfe7b472**827dce9cc6..**
> ac94a009111cf230ef70984add4b22**034f53fdac 100644
> --- a/base/lazy_instance.h
> +++ b/base/lazy_instance.h
> @@ -133,8 +133,11 @@ class LazyInstance : public LazyInstanceHelper {
>   }
>
>   Type* Pointer() {
> +#ifndef NDEBUG
> +    // Avoid making TLS lookup on release builds.
>     if (!Traits::**kAllowedToAccessOnNonjoinableT**hread)
>       base::ThreadRestrictions::**AssertSingletonAllowed();
> +#endif
>
>     // We will hopefully have fast access when the instance is already
> created.
>     // Since a thread sees state_ != STATE_CREATED at most once,
> Index: base/memory/singleton.h
> diff --git a/base/memory/singleton.h b/base/memory/singleton.h
> index bc4684b0b8940acf87ca0163414657**64967a26ab..**
> f82a37ec865da1a966cb3b3681d2d1**edf4ba1427 100644
> --- a/base/memory/singleton.h
> +++ b/base/memory/singleton.h
> @@ -213,8 +213,11 @@ class Singleton {
>
>   // Return a pointer to the one true instance of the class.
>   static Type* get() {
> +#ifndef NDEBUG
> +    // Avoid making TLS lookup on release builds.
>     if (!Traits::**kAllowedToAccessOnNonjoinableT**hread)
>       base::ThreadRestrictions::**AssertSingletonAllowed();
> +#endif
>
>     // Our AtomicWord doubles as a spinlock, where a value of
>     // kBeingCreatedMarker means the spinlock is being held for creation.
>
>
>

[commit-bot: I haz the power, 2011-11-04 17:03:58]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/joth@chromium.org/8465010/1

[commit-bot: I haz the power, 2011-11-04 17:04:00]

Presubmit check for 8465010-1 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Found Singleton<T> in the following header files.
Please move them to an appropriate source file so that the template gets
instantiated in a single compilation unit.
  base/memory/singleton.h

Was the presubmit check useful? Please send feedback & hate mail to
maruel@chromium.org!

[willchan no longer on Chromium, 2011-11-04 17:14:58]

Haha, so pwn3d. Bypass this presubmit please :)

On Fri, Nov 4, 2011 at 10:04 AM, <commit-bot@chromium.org> wrote:

> Presubmit check for 8465010-1 failed and returned exit status 1.
>
> Running presubmit commit checks ...
>
> ** Presubmit ERRORS **
> Found Singleton<T> in the following header files.
> Please move them to an appropriate source file so that the template gets
> instantiated in a single compilation unit.
>  base/memory/singleton.h
>
> Was the presubmit check useful? Please send feedback & hate mail to
> maruel@chromium.org!
>
>
>
http://codereview.chromium.**org/8465010/<http://codereview.chromium.org/8465...
>