Consider pmemrange regions when trying to free pages from the inactive list.

Instead of starting with the first page on the inactive list pick the first
one that fits in the range of a given pmemrange region.

Fix an issue where the page daemon would be hogging a CPU without freeing any
page because the global limits are satisfied.

The algorithm could certainly be revisited, the LRU is not really working and
there is a huge delay before recovering when the first pages need to be swapped and the system is in OOM situation but at least there is no starvation and the
system now recovers.

ok beck@, kettenis@

1 files changed, 31 insertions, 11 deletions

diff --git a/sys/uvm/uvm_pdaemon.c b/sys/uvm/uvm_pdaemon.c
index 99cadaa06de..52baa2eb590 100644
--- a/sys/uvm/uvm_pdaemon.c
+++ b/sys/uvm/uvm_pdaemon.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: uvm_pdaemon.c,v 1.100 2022/06/28 19:23:08 mpi Exp $	*/
+/*	$OpenBSD: uvm_pdaemon.c,v 1.101 2022/06/28 19:31:30 mpi Exp $	*/
 /*	$NetBSD: uvm_pdaemon.c,v 1.23 2000/08/20 10:24:14 bjh21 Exp $	*/
 
 /*
@@ -101,8 +101,8 @@ extern void drmbackoff(long);
  * local prototypes
  */
 
-void		uvmpd_scan(void);
-boolean_t	uvmpd_scan_inactive(struct pglist *);
+void		uvmpd_scan(struct uvm_pmalloc *);
+boolean_t	uvmpd_scan_inactive(struct uvm_pmalloc *, struct pglist *);
 void		uvmpd_tune(void);
 void		uvmpd_drop(struct pglist *);
 
@@ -281,7 +281,7 @@ uvm_pageout(void *arg)
 		if (pma != NULL ||
 		    ((uvmexp.free - BUFPAGES_DEFICIT) < uvmexp.freetarg) ||
 		    ((uvmexp.inactive + BUFPAGES_INACT) < uvmexp.inactarg)) {
-			uvmpd_scan();
+			uvmpd_scan(pma);
 		}
 
 		/*
@@ -379,7 +379,7 @@ uvm_aiodone_daemon(void *arg)
  */
 
 boolean_t
-uvmpd_scan_inactive(struct pglist *pglst)
+uvmpd_scan_inactive(struct uvm_pmalloc *pma, struct pglist *pglst)
 {
 	boolean_t retval = FALSE;	/* assume we haven't hit target */
 	int free, result;
@@ -404,8 +404,27 @@ uvmpd_scan_inactive(struct pglist *pglst)
 	swnpages = swcpages = 0;
 	free = 0;
 	dirtyreacts = 0;
+	p = NULL;
 
-	for (p = TAILQ_FIRST(pglst); p != NULL || swslot != 0; p = nextpg) {
+	/* Start with the first page on the list that fit in pma's ranges */
+	if (pma != NULL) {
+		paddr_t paddr;
+
+		TAILQ_FOREACH(p, pglst, pageq) {
+			paddr = atop(VM_PAGE_TO_PHYS(p));
+			if (paddr >= pma->pm_constraint.ucr_low &&
+			    paddr < pma->pm_constraint.ucr_high)
+				break;
+		}
+
+	}
+
+	if (p == NULL) {
+		p = TAILQ_FIRST(pglst);
+		pma = NULL;
+	}
+
+	for (; p != NULL || swslot != 0; p = nextpg) {
 		/*
 		 * note that p can be NULL iff we have traversed the whole
 		 * list and need to do one final swap-backed clustered pageout.
@@ -419,8 +438,8 @@ uvmpd_scan_inactive(struct pglist *pglst)
 			 * our target
 			 */
 			free = uvmexp.free - BUFPAGES_DEFICIT;
-
-			if (free + uvmexp.paging >= uvmexp.freetarg << 2 ||
+			if (((pma == NULL || (pma->pm_flags & UVM_PMA_FREED)) &&
+			    (free + uvmexp.paging >= uvmexp.freetarg << 2)) ||
 			    dirtyreacts == UVMPD_NUMDIRTYREACTS) {
 				retval = TRUE;
 
@@ -531,7 +550,8 @@ uvmpd_scan_inactive(struct pglist *pglst)
 			 * this page is dirty, skip it if we'll have met our
 			 * free target when all the current pageouts complete.
 			 */
-			if (free + uvmexp.paging > uvmexp.freetarg << 2) {
+			if ((pma == NULL || (pma->pm_flags & UVM_PMA_FREED)) &&
+			    (free + uvmexp.paging > uvmexp.freetarg << 2)) {
 				if (anon) {
 					rw_exit(anon->an_lock);
 				} else {
@@ -867,7 +887,7 @@ uvmpd_scan_inactive(struct pglist *pglst)
  */
 
 void
-uvmpd_scan(void)
+uvmpd_scan(struct uvm_pmalloc *pma)
 {
 	int free, inactive_shortage, swap_shortage, pages_freed;
 	struct vm_page *p, *nextpg;
@@ -910,7 +930,7 @@ uvmpd_scan(void)
 	 * low bit of uvmexp.pdrevs (which we bump by one each call).
 	 */
 	pages_freed = uvmexp.pdfreed;
-	(void) uvmpd_scan_inactive(&uvm.page_inactive);
+	(void) uvmpd_scan_inactive(pma, &uvm.page_inactive);
 	pages_freed = uvmexp.pdfreed - pages_freed;
 
 	/*