Thanks. When I first did the thread plan stuff I wasn't sure whether I would
have a use for an empty plan stack, so I didn't rigorously go through and
assert everywhere that might happen. I had some idea for using an empty plan
stack which was apparently so kooky that not only did I never impl
I'll change it to an assert and next time it happens I'll dig a little
more. I've only seen this on about 1 or 2 out of 100 runs of the test
suite so I don't know how it happens.
On Fri, Nov 13, 2015 at 1:44 PM Jim Ingham wrote:
> GetCurrentPlan should never return NULL. The Thread constructor
GetCurrentPlan should never return NULL. The Thread constructor pushes the
Base thread plan onto the stack, and pop won't remove the Base plan. When we
destroy the thread (in ThreadDestroy) we clear the plan stack and then push a
ThreadPlanNull onto the plan stack. So if you are seeing a GetC
Author: zturner
Date: Fri Nov 13 15:28:53 2015
New Revision: 253086
URL: http://llvm.org/viewvc/llvm-project?rev=253086&view=rev
Log:
Add a null check against the ThreadPlan
I'm seeing some cases where the ThreadPlan is null. It could
be a sign of a valid race condition, but at least we shouldn'
