Javier Delgadillo wrote:
> Were you able to determine which other CMT_* functions were holding the
> lock at around the same time?
While I'm sat waiting for something in EventLoop I'm seeing activity on the
socket from another thread. That other thread is nsSocketTransport::IsAlive
which does a zero length read. This zero length read gets us into
nsSSLIOLayerRead where, because the read didn't actually read anything, we
get into the error code where we send a message to the PSM message. Is this
expected? Shouldn't the zero status coming back from read/recv when amount
is zero (in nsSSLIOLayerRead) be treated as a success? I tried changing it
by now I'm hanging.
> I'm not sure why this isn't an issue on other platforms, perhaps there's
> a bug in the PRMonitor on OpenVMS.
PRMonitor must be working otherwise all sorts of other stuff would be
failing. I think its timing related.
Colin.
