kernel/vm: Add better handling for busy-writing pages on cache resizes.

Previously, busy-writing pages just stayed in the cache until
I/O finished. This wasn't right, as they should've been unmapped
(and removed) immediately, as a TODO in the PageWriteWrapper noted.

Now they are, and they're moved into a queue on the cache pending
the I/O completing.

This paves the way for fixing #18390.

Change-Id: I51778ba1cfc1d26f34dd44f1ee293381be59937f
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8627
Reviewed-by: waddlesplash <[email protected]>
Tested-by: Commit checker robot <[email protected]>

headers/private/kernel/vm/VMCache.h

@@ -115,6 +115,7 @@ public:
 									{ if (fPageEventWaiters != NULL)
 										_NotifyPageEvents(page, events); }
 	inline	void				MarkPageUnbusy(vm_page* page);
+			void				FreeRemovedPage(vm_page* page);
 
 			vm_page*			LookupPage(off_t offset);
 			void				InsertPage(vm_page* page, off_t offset);
@@ -231,12 +232,17 @@ private:
 			bool				_FreePageRange(VMCachePagesTree::Iterator it,
 									page_num_t* toPage, page_num_t* freedPages);
 
+private:
+	typedef DoublyLinkedQueue<vm_page, DoublyLinkedListCLink<vm_page,
+		SplayTreeLink<vm_page>, &vm_page::cache_link> > RemovedPagesQueue;
+
 private:
 			int32				fRefCount;
 			mutex				fLock;
 			PageEventWaiter*	fPageEventWaiters;
 			void*				fUserData;
 			VMCacheRef*			fCacheRef;
+			RemovedPagesQueue	fRemovedBusyPages;
 
 			page_num_t			fWiredPagesCount;
 			uint64				fFaultCount;