Allow the VM object flushing code to cluster. When the filesystem syncer

comes along and flushes a file which has been mmap()'d SHARED/RW, with
dirty pages, it was flushing the underlying VM object asynchronously,
resulting in thousands of 8K writes.  With this change the VM Object flushing
code will cluster dirty pages in 64K blocks.

Note that until the low memory deadlock issue is reviewed, it is not safe
to allow the pageout daemon to use this feature.  Forced pageouts still
use fs block size'd ops for the moment.

MFC after:	3 days

3 files changed, 22 insertions, 7 deletions

diff --git a/sys/vm/vm_object.c b/sys/vm/vm_object.c
index 33f9d67..d9cd47f 100644
--- a/sys/vm/vm_object.c
+++ b/sys/vm/vm_object.c
@@ -631,6 +631,9 @@ vm_object_terminate(vm_object_t object)
  *	write out pages with PG_NOSYNC set (originally comes from MAP_NOSYNC),
  *	leaving the object dirty.
  *
+ *	When stuffing pages asynchronously, allow clustering.  XXX we need a
+ *	synchronous clustering mode implementation.
+ *
  *	Odd semantics: if start == end, we clean everything.
  *
  *	The object must be locked.
@@ -652,7 +655,7 @@ vm_object_page_clean(vm_object_t object, vm_pindex_t start, vm_pindex_t end, int
 		(object->flags & OBJ_MIGHTBEDIRTY) == 0)
 		return;
 
-	pagerflags = (flags & (OBJPC_SYNC | OBJPC_INVAL)) ? VM_PAGER_PUT_SYNC : 0;
+	pagerflags = (flags & (OBJPC_SYNC | OBJPC_INVAL)) ? VM_PAGER_PUT_SYNC : VM_PAGER_CLUSTER_OK;
 	pagerflags |= (flags & OBJPC_INVAL) ? VM_PAGER_PUT_INVAL : 0;
 
 	vp = object->handle;
@@ -682,6 +685,7 @@ vm_object_page_clean(vm_object_t object, vm_pindex_t start, vm_pindex_t end, int
 		scanreset = object->resident_page_count / EASY_SCAN_FACTOR;
 		if (scanreset < 16)
 			scanreset = 16;
+		pagerflags |= VM_PAGER_IGNORE_CLEANCHK;
 
 		scanlimit = scanreset;
 		tscan = tstart;
@@ -732,6 +736,7 @@ vm_object_page_clean(vm_object_t object, vm_pindex_t start, vm_pindex_t end, int
 			vm_object_clear_flag(object, OBJ_CLEANING);
 			return;
 		}
+		pagerflags &= ~VM_PAGER_IGNORE_CLEANCHK;
 	}
 
 	/*
@@ -854,7 +859,8 @@ vm_object_page_collect_flush(vm_object_t object, vm_page_t p, int curgeneration,
 
 		if ((tp = vm_page_lookup(object, pi + i)) != NULL) {
 			if ((tp->flags & PG_BUSY) ||
-				(tp->flags & PG_CLEANCHK) == 0 ||
+				((pagerflags & VM_PAGER_IGNORE_CLEANCHK) == 0 &&
+				 (tp->flags & PG_CLEANCHK) == 0) ||
 				(tp->busy != 0))
 				break;
 			if((tp->queue - tp->pc) == PQ_CACHE) {
@@ -881,7 +887,8 @@ vm_object_page_collect_flush(vm_object_t object, vm_page_t p, int curgeneration,
 
 			if ((tp = vm_page_lookup(object, pi - i)) != NULL) {
 				if ((tp->flags & PG_BUSY) ||
-					(tp->flags & PG_CLEANCHK) == 0 ||
+					((pagerflags & VM_PAGER_IGNORE_CLEANCHK) == 0 &&
+					 (tp->flags & PG_CLEANCHK) == 0) ||
 					(tp->busy != 0))
 					break;
 				if ((tp->queue - tp->pc) == PQ_CACHE) {
diff --git a/sys/vm/vm_pager.h b/sys/vm/vm_pager.h
index 2c6d196..07b6878 100644
--- a/sys/vm/vm_pager.h
+++ b/sys/vm/vm_pager.h
@@ -79,8 +79,10 @@ struct pagerops {
 #define	VM_PAGER_ERROR	4
 #define VM_PAGER_AGAIN	5
 
-#define	VM_PAGER_PUT_SYNC	0x1
-#define	VM_PAGER_PUT_INVAL	0x2
+#define	VM_PAGER_PUT_SYNC		0x0001
+#define	VM_PAGER_PUT_INVAL		0x0002
+#define VM_PAGER_IGNORE_CLEANCHK	0x0004
+#define VM_PAGER_CLUSTER_OK		0x0008
 
 #ifdef _KERNEL
 #ifdef MALLOC_DECLARE
diff --git a/sys/vm/vnode_pager.c b/sys/vm/vnode_pager.c
index cf98090..667bd3a 100644
--- a/sys/vm/vnode_pager.c
+++ b/sys/vm/vnode_pager.c
@@ -1044,11 +1044,17 @@ vnode_pager_generic_putpages(vp, m, bytecount, flags, rtvals)
 	/*
 	 * pageouts are already clustered, use IO_ASYNC t o force a bawrite()
 	 * rather then a bdwrite() to prevent paging I/O from saturating 
-	 * the buffer cache.
+	 * the buffer cache.  Dummy-up the sequential heuristic to cause
+	 * large ranges to cluster.  If neither IO_SYNC or IO_ASYNC is set,
+	 * the system decides how to cluster.
 	 */
 	ioflags = IO_VMIO;
-	ioflags |= (flags & (VM_PAGER_PUT_SYNC | VM_PAGER_PUT_INVAL)) ? IO_SYNC: IO_ASYNC;
+	if (flags & (VM_PAGER_PUT_SYNC | VM_PAGER_PUT_INVAL))
+		ioflags |= IO_SYNC;
+	else if ((flags & VM_PAGER_CLUSTER_OK) == 0)
+		ioflags |= IO_ASYNC;
 	ioflags |= (flags & VM_PAGER_PUT_INVAL) ? IO_INVAL: 0;
+	ioflags |= IO_SEQMAX << IO_SEQSHIFT;
 
 	aiov.iov_base = (caddr_t) 0;
 	aiov.iov_len = maxsize;