linux: call PR_Init on UI thread

linux: call PR_Init on UI thread

It complains about being shutdown on the wrong thread otherwise.

BUG=18410

[Evan Martin, 2009-09-01 21:31:57]

This is kind of a lame fix, but maybe you'll accept it

[wtc, 2009-09-01 22:13:35]

LGTM.

Another solution is to add an EnsureNSPRInit function and
call it in BrowserMain.  Call PR_Init to initialize NSPR
explicitly:
https://developer.mozilla.org/en/PR_Init

http://codereview.chromium.org/178062/diff/4001/5001
File base/nss_init.cc (right):

http://codereview.chromium.org/178062/diff/4001/5001#newcode146
Line 146: // otherwise, normal process cleanup will do this work for us.
Would be nice to add a comment that says PR_Cleanup fails if
NSPR isn't initialized and cleaned up on the same thread.

[Evan Martin, 2009-09-02 00:57:08]

Please look, reimplemented based on conversation with WTC.

[wtc, 2009-09-02 01:17:17]

You can check this in after fixing the issues below.
Please upload a new Patch Set before you check it in,
so I can doublecheck it after the fact.  Thanks!

http://codereview.chromium.org/178062/diff/8001/8002
File base/nss_init.cc (right):

http://codereview.chromium.org/178062/diff/8001/8002#newcode1
Line 1: // Copyright (c) 2009 The Chromium Authors. All rights reserved.
This should be 2008-2009.

http://codereview.chromium.org/178062/diff/8001/8002#newcode60
Line 60: // Separate from the NSS singleton because we initialize it on the UI
thread.
Nit: it => NSPR (otherwise it's not clear what "it" refers to)

Let's change "UI thread" to "main thread".

http://codereview.chromium.org/178062/diff/8001/8002#newcode71
Line 71: "Is NSPR getting destroyed on wrong thread?";
Since Singleton destructors run on the main thread, the
question should be "Is NSPR getting initialized on the
wrong thread?" or "Was NSPR initialized on the wrong thread?"
(not sure about the tense/grammar :-))

http://codereview.chromium.org/178062/diff/8001/8002#newcode79
Line 79: SECStatus status;
We should call EnsureNSPRInit here, so that the unit tests,
etc. will continue to call PR_Cleanup.

http://codereview.chromium.org/178062/diff/8001/8003
File base/nss_init.h (right):

http://codereview.chromium.org/178062/diff/8001/8003#newcode1
Line 1: // Copyright (c) 2009 The Chromium Authors. All rights reserved.
This should say 2008-2009.

http://codereview.chromium.org/178062/diff/8001/8004
File base/singleton.h (right):

http://codereview.chromium.org/178062/diff/8001/8004#newcode1
Line 1: // Copyright (c) 2009 The Chromium Authors. All rights reserved.
This should say 2006-2009.

http://codereview.chromium.org/178062/diff/8001/8005
File chrome/app/chrome_dll_main.cc (right):

http://codereview.chromium.org/178062/diff/8001/8005#newcode1
Line 1: // Copyright (c) 2009 The Chromium Authors. All rights reserved.
Should be 2006-2009.

http://codereview.chromium.org/178062/diff/8001/8005#newcode553
Line 553: base::EnsureNSPRInit();
Are we using NSS/NSPR in the renderer processes, too?
If not, we just need to call EnsureNSPRInit in BrowserMain.

We may need to call EnsureNSPRInit in the test shell...