Ever since February 2010, the body of the login coroutine function has been
enclosed in try/catch (...), with an llerrs message to try to crash more
informatively than the runtime's unhandled-exception termination. Over the
years this evolved to LL_ERRS and then to CRASH_ON_UNHANDLED_EXCEPTION.
This persisted despite the August 2016 addition of generic catch clauses in
the LLCoros::toplevel() function to serve the same purpose, and despite the
subsequent introduction of the LLCoros::Stop family of exceptions to
deliberately throw into waiting coroutines on viewer shutdown.
That's exactly what was happening. When the user closed the viewer while
waiting for the response from login.cgi, the waiting operation threw
LLCoros::Stopping, which was caught by that CRASH_ON_UNHANDLED_EXCEPTION,
which crashed the viewer with LL_ERRS rather than propagating up to the
toplevel() and cleanly terminating the coroutine.
Change CRASH_ON_UNHANDLED_EXCEPTION() to LOG_UNHANDLED_EXCEPTION() and
re-throw so toplevel() can handle.
The llappviewerwin32.cpp create_console() function called by
LLAppViewerWin32::initConsole() used to assign *stderr = *(new FILE* value),
and so forth for stdout and stdin. That dubious tactic no longer works with
the new Windows CRT introduced with VS 2015. freopen_s() works much better.
There are separate stanzas in llappviewerwin32.cpp's create_console() function
for each of STD_INPUT_HANDLE, STD_OUTPUT_HANDLE and STD_ERROR_HANDLE. SL-13361
wants to add more code to each. Factor out new local set_stream() function and
make create_console() call it three times.
(cherry picked from commit 13b78a0c5a)
Specifically, llcoro::suspendUntilTimeout() is definitely called concurrently
by multiple coroutines. New code that instantiates a local LLEventStream must
allow the name to be tweaked for uniqueness.
Specifically, the shutdown crash reported in SL-13512 was due to
LLExperienceCache::idleCoro() looping on suspendUntilTimeout(), failing to
notice in its slumbers that the viewer was shutting down around it.
Make suspendUntilTimeout() internally call suspendUntilEventOnWithTimeout(),
which already listens for "LLApp" state-change events and throws Stopping when
LLApp enters its shutdown sequence.
There are duplicate LLTrace::BlockTimerStatHandle key strings declared in
llsettingsdaycycle.cpp and llsettingswater.cpp -- the only instances of
duplicate BlockTimerStatHandle keys in the viewer code base. SL-13467 tracks
intentional crashes due to duplicate LLInstanceTracker subclass instances with
one of those keys. The simplest experiment to try to eliminate those crashes
is to ensure that every BlockTimerStatHandle in the code base is unique.
The observed failure is that SLVoice, on relaunch, produces an error that
bind() returned EADDRINUSE and terminates. Using a different port every time
we relaunch avoids that collision.
Enable the body of the existing ll_debug_socket() function (on Mac as well as
Linux), but using tag "Socket" so you can turn on its log messages without
emitting *all* debug messages.
It can happen that we arrive at logoutOfVivox() with some other message
queued on the LLEventMailDrop in question. If logoutOfVivox() assumes that
other message is logout and exits, then subsequent code gets confused.
Introduce a loop to wait (with the existing timeout) for the real logout
message.
Kitty Barnett