Twice in the last few days we've seen the following crash on the bots.

[32268:125992:0728/123353:391400844:FATAL:time_win.cc(341)] Check failed: now - ReliableNow() - skew_ < kMaxTimeDrift. 
Backtrace:
	StackTrace::StackTrace [0x00B44141+33] (c:\b\slave\sub-dbg-xp\build\src\base\debug_util_win.cc:230)
	logging::LogMessage::~LogMessage [0x00AE66B1+81] (c:\b\slave\sub-dbg-xp\build\src\base\logging.cc:465)
	`anonymous namespace'::HighResNowSingleton::Now [0x00AE51B3+243] (c:\b\slave\sub-dbg-xp\build\src\base\time_win.cc:343)
	base::TimeTicks::HighResNow [0x00AE5087+39] (c:\b\slave\sub-dbg-xp\build\src\base\time_win.cc:399)
	net::BandwidthMetrics::StartStream [0x00BF8CFC+60] (c:\b\slave\sub-dbg-xp\build\src\net\base\bandwidth_metrics.h:74)
	net::ScopedBandwidthMetrics::StartStream [0x00BF8CAF+31] (c:\b\slave\sub-dbg-xp\build\src\net\base\bandwidth_metrics.h:136)
	net::SpdyStream::OnResponseReceived [0x00BF8A7A+218] (c:\b\slave\sub-dbg-xp\build\src\net\spdy\spdy_stream.cc:123)

I'm tentatively giving this to Mike since he understands the intersection of {windows, timers, spdy}.

The test is
	net::SpdyNetworkTransactionTest_Post_Test::TestBody [0x0043DFFF+591] (c:\b\slave\sub-dbg-xp\build\src\net\spdy\spdy_network_transaction_unittest.cc:817)
maybe that is fiddling with timers somehow?