-
-/* Should we filter stack/register pointers? This substantially reduces the
- * number of invalid pointers accepted.
- *
- * FIXME: This is basically constant=1. It will probably degrade
- * interrupt safety during object initialization. But I don't think we
- * should do without it -- the possibility of the GC being too
- * conservative and hence running out of memory is also. Perhaps the
- * interrupt safety issue could be fixed by making the initialization
- * code do WITHOUT-GCING or WITHOUT-INTERRUPTS until the appropriate
- * type bits have been set. (That might be necessary anyway, in order
- * to keep interrupt code's allocation operations from stepping on the
- * interrupted code's allocations.) Or perhaps it could be fixed by
- * making sure that uninitialized memory is zero, reserving the
- * all-zero case for uninitialized memory, and making the
- * is-it-possibly-a-valid-pointer code check for all-zero and return
- * true in that case. Then after either fix, we could get rid of this
- * variable and simply hardwire the system always to do pointer
- * filtering. */
-boolean enable_pointer_filter = 1;