After analyzing some crash dumps I saw a race condition in the governor. A helper thread referencing a global event was in a race condition with another thread that initialized the used handle at a later time (a minor code positioning change where the impact wasn't seen). At least I believe this to be the case. I'm still having some troubles getting my debug symbols archived and matched properly, something I'll figure out soon enough. The crash location and disassembly was pretty compelling though, and this would explain the seemingly random occurrences of this (mostly all reported under x32 platforms).
Otherwise I haven't added anything new, yet. I do have some last minute additions planned. Simple things that will make a big difference. First I want to get this QA out of the way.
Tuesday, 15 March 2011
Subscribe to:
Post Comments (Atom)
0 comments:
Post a Comment