The crash resolves in hash table code.
When DRT calls WebKitCreateInstance it tries to populate a hash table. At some point it needs to expand the table and reinsert objects into the new larger table but for some reason one of the objects it tries to reinsert has been marked as deleted and things go downhill from there.
This problem only reproduces locally on my Windows machine. The Apple Windows port test bots don't seem to have the same issue.
Originally, there were crashes occuring after r128583. These were fixed later by a workaround and then an actual fix in r128850.
I failed to notice this fix while I was bisecting but I believe that the crashing form r128583 and the ones occuring now are because of different reasons. Most likely soething after r 128850 made things crash again.
I'm going to investigate r128850 to see whether or not the crashes still occur at that revision.
It crashes on r128850 as well.
In addition. I tried running the tests locally on the bots. It also crashes there an generates an error ::LoadLibraryW failed (whatever that means).
Anyways, this means that it is not a problem specific to my setup but simply running tests locally in general.
I may just have to rely on webkit test runner but this is definitely an issue we need to solve.
I'll run DRT from VS2005 and print out the crashing stack trace. I doubt it will be particularly helpful though.
What about running r128582 or older locally?
everything runs fine before r128583, i'm now running it on r128629, the workaround patch
o hey, it doesn't crash there...there's a decent chance that the crash actually has nothing to do at all with the revisions I mentioned earlier. I'll keep regression testing.