Commit dddfba38 authored by Simon Marlow's avatar Simon Marlow
Browse files

Fix another subtle shutdown deadlock

The problem occurred when a thread tries to GC during shutdown.  In
order to GC it has to acquire all the Capabilities in the system, but
during shutdown, some of the Capabilities have already been closed and
can never be acquired.
parent 6d316585
......@@ -340,7 +340,14 @@ schedule (Capability *initialCapability, Task *task)
/* scheduleDoGC() deletes all the threads */
cap = scheduleDoGC(cap,task,rtsFalse);
// after scheduleDoGC(), we must be shutting down. Either some
// other Capability did the final GC, or we did it above,
// either way we can fall through to the SCHED_SHUTTING_DOWN
// case now.
// fall through
debugTrace(DEBUG_sched, "SCHED_SHUTTING_DOWN");
// If we are a worker, just exit. If we're a bound thread
......@@ -1467,6 +1474,13 @@ scheduleDoGC (Capability *cap, Task *task USED_IF_THREADS, rtsBool force_major)
nat i;
if (sched_state == SCHED_SHUTTING_DOWN) {
// The final GC has already been done, and the system is
// shutting down. We'll probably deadlock if we try to GC
// now.
return cap;
// In order to GC, there must be no threads running Haskell code.
// Therefore, the GC thread needs to hold *all* the capabilities,
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment