Handle rare conditions, like message queue exhaustion

Handle rare conditions, like message queue exhaustion

bug=128734
r=rvargas

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=138063

[rvargas (doing something else), 2012-05-19 00:19:10]

lgtm

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win.cc
File base/message_pump_win.cc (right):

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win....
base/message_pump_win.cc:126: // will probably be workable.
I cannot say don't comment the code... but the other extreme is not good either.
For instance everything after "we'll put the have_work_ flag" is actually
redundant with the code... and the comment on that line of code!

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win....
base/message_pump_win.cc:164: CHECK(ret);  // If we can't set timers, we are in
big trouble.
I'd say do a GetLastError before crashing... and be prepared for an increased
crash rate.

[jar (doing other things), 2012-05-19 01:01:19]

Changed to more conservative counting of previousyl ignored SetTimer error,
rather than crashing.... and we'll see what we should do when the histogram
counter comes in with real user data.

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win.cc
File base/message_pump_win.cc (right):

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win....
base/message_pump_win.cc:126: // will probably be workable.
I shortened the more redundant section..

On 2012/05/19 00:19:11, rvargas wrote:
> I cannot say don't comment the code... but the other extreme is not good
either.
> For instance everything after "we'll put the have_work_ flag" is actually
> redundant with the code... and the comment on that line of code!

https://chromiumcodereview.appspot.com/10392163/diff/1/base/message_pump_win....
base/message_pump_win.cc:164: CHECK(ret);  // If we can't set timers, we are in
big trouble.
I switched to histogramming the presence of this error, rather than crashing. 
IF we don't see the histogram tallying, we can then try a CHECK() instead.

On 2012/05/19 00:19:11, rvargas wrote:
> I'd say do a GetLastError before crashing... and be prepared for an increased
> crash rate.

[rvargas (doing something else), 2012-05-19 01:23:04]

lgtm

[commit-bot: I haz the power, 2012-05-20 15:05:13]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jar@chromium.org/10392163/4001

[commit-bot: I haz the power, 2012-05-20 16:28:52]

Change committed as 138063