[Bug 16051] LOTRO launcher fails on startup

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=16051


Jeremy White <jwhite at codeweavers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jwhite at codeweavers.com




--- Comment #8 from Jeremy White <jwhite at codeweavers.com>  2009-01-03 09:57:13 ---
I've spent some time on this, and have a few bits of further information.

First, in addition to the 'Fatal Execution Error' and the 'Failed to load
resources' error that are very common, you also get a divide by zero error at
times.  This appears to be related to
NtQuerySystemInformation(SystemProcessorPerformanceInformation,...). 
Specifically, you see a thread get the sspi, Sleep(500), and get the sspi and
then a divide by zero.  I've run some tests on my Linux boxen, and in fact,
I'll see the usr clock ticks not change sometimes.  So it seems to be a Linux
-> Windows difference, in that a Sleep(500) guarantees the user time will
change on Windows, but apparently not on Linux.  I haven't probed much further,
but I'll attach the clockbias hack I used to prevent this from happening.

Second, there seems to be a SEGV that triggers a stack recursion that
eventually blows the stack.  That is, I'll see a RaiseException(c0000005,...),
followed by exception handling code (you can see it looking for 'EHGolden'),
and then exception handling code loops again and again, and then the stack
blows.  I believe this is what triggers the Fatal Execution Error, but the
threads are sufficiently disconnected that I may be wrong.

You can fudge that if you prevent the RSS and launcher thread connect()s from
firing.  That actually gets the launcher to go much further and even start
updating files.  I've attached a hack for that as well.  Of course, with that,
you go on to fail after downloading for perhaps 30-40 seconds.  Start it up,
you get 30-40 more seconds of downloads, and then another crash.  I haven't
probed those further failures, they may provide some clues.

There also seems to clearly be just some flat out instability in running .NET
1.1.  If you install the .net installer, you'll sometimes get a 'failed to load
resources' error.  And, looking at normal wine fixmes from installing .NET 1.1
and it's service pack, it sure smells like it does not fully install.

p.s. Note that you need Alexandre's server crash fix as of January 3rd, or
you'll have the wine server dropping core on you.


-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.