kernel/vm: Make cut_area respect overcommitting flag

`cut_area` now checks the protection of the target area for
the `B_OVERCOMMITTING_AREA` flag and sets it on any cache it
creates. This allows operations that split large overcommitting
areas to succeed.

Change-Id: I8dee27d22df29741cc61af2575c9e6626da25949
Reviewed-on: https://review.haiku-os.org/c/haiku/+/6391
Tested-by: Commit checker robot <[email protected]>
Reviewed-by: waddlesplash <[email protected]>

src/system/kernel/vm/vm.cpp

@@ -745,7 +745,8 @@ cut_area(VMAddressSpace* addressSpace, VMArea* area, addr_t address,
 	if (onlyCacheUser) {
 		// Create a new cache for the second area.
 		VMCache* secondCache;
-		error = VMCacheFactory::CreateAnonymousCache(secondCache, false, 0, 0,
+		error = VMCacheFactory::CreateAnonymousCache(secondCache,
+			area->protection & B_OVERCOMMITTING_AREA, 0, 0,
 			dynamic_cast<VMAnonymousNoSwapCache*>(cache) == NULL, priority);
 		if (error != B_OK) {
 			addressSpace->ShrinkAreaTail(area, oldSize, allocationFlags);