Make it possible to use ThreadLocalStorage::Slot as a static without

Make it possible to use ThreadLocalStorage::Slot as a static without
introducing static initializers.

Member variables can stay TLS::Slots and their behavior doesn't change.
Static instances use TLS::StaticSlot instead.

Kind of like http://codereview.chromium.org/8491043 but for TLS.

BUG=none
TEST=none
TBR=agl

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=119679

[Nico, 2012-01-27 06:24:38]

This is currently more proof-of-concept than a real CL. If you don't hate it,
I'll clean it up (and check that it actually helps).

[brettw, 2012-01-27 20:52:14]

Seems like a fine approach. I wonder if TLS_INITIALIZER should be
BASE_TLS_INITIALIZER for some scoping, or do you think that looks stupid? It's
probably OK either way.

[Nico, 2012-01-28 03:37:10]

I verified that this does indeed remove 3 static initializers on linux (which
wasn't obvious since tools/linux/dump-static-initializers.py doesn't print these
3 for some reason). PTAL.

We don't use the BASE_ prefix for LINKER_INITIALIZED and
LAZY_INSTANCE_INITIALIZER, so I didn't use it here either. If you want, I can
add it, I'm fine either way.

evan: fyi. Once, for the d-s-i.py information, two because you complained about
us having LazyInstances for ThreadLocalPointers – we can probably use this same
technique to get rid of them.

[brettw, 2012-01-28 03:41:41]

Actually, we use base::LINKER_INITIALIZED which is even better :) because it's
an enum. Though for this case I think it has to be a macro so we're stuck. I
guess what you have is fine, lemme look at the patch in a bit.

[Evan Martin, 2012-01-28 18:12:10]

On 2012/01/28 03:37:10, Nico wrote:
> I verified that this does indeed remove 3 static initializers on linux (which
> wasn't obvious since tools/linux/dump-static-initializers.py doesn't print
these
> 3 for some reason). PTAL.

Is it due to the two-byte ctor thing that d-s-i filters out?

[brettw, 2012-01-30 05:46:17]

lgtm

[commit-bot: I haz the power, 2012-01-30 05:47:34]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/9297010/2008

[commit-bot: I haz the power, 2012-01-30 05:47:38]

Presubmit check for 9297010-2008 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for: net/base/dns_reloader.cc

Presubmit checks took 1.2s to calculate.

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

[Nico, 2012-01-30 05:48:39]

+agl for net OWNERShip

[commit-bot: I haz the power, 2012-01-30 05:48:54]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/9297010/2008

[commit-bot: I haz the power, 2012-01-30 07:38:36]

Try job failure for 9297010-2008 (retry) on mac_rel for step "net_unittests".
It's a second try, previously, step "net_unittests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[commit-bot: I haz the power, 2012-01-30 07:40:08]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/9297010/2008

[commit-bot: I haz the power, 2012-01-30 07:41:52]

Try job failure for 9297010-2008 on linux_rel for step "update".
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...

Step "update" is always a major failure.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2012-01-30 15:43:03]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/9297010/2008

[agl, 2012-01-30 16:04:36]

LGTM for net/