Christophe Rhodes <csr21 <at> cam.ac.uk> writes:
> called once and then hanging. That would be the difference between
> something going wrong with the spinlock in with-pcl-lock, and
> something going wrong elsewhere. Try tracing
> sb-pcl::flush-cache-vector-internal just before you call the thing
> that hangs.
Clearly with-spin-lock hangs, because it gets called twice recursively. From
force-flush-caches and from flush-cache-vector-internal.
Moving (setf (slot-value class 'wrapper) nwrapper) outside the with-pcl-lock
part of force-cache-flushes is a workaround for my problem although it is
probably an incorrect solution.