Skip site navigation (1)Skip section navigation (2) 
Date:      Sun, 25 Apr 2010 20:40:45 +0000 (UTC)
From:      Alan Cox <alc@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   svn commit: r207205 - in head/sys: amd64/amd64 i386/i386
Message-ID:  <201004252040.o3PKejOe065064@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help 
Author: alc
Date: Sun Apr 25 20:40:45 2010
New Revision: 207205
URL: http://svn.freebsd.org/changeset/base/207205

Log:
  Clearing a page table entry's accessed bit (PG_A) and setting the
  page's PG_REFERENCED flag in pmap_protect() can't really be justified.
  In contrast to pmap_remove() or pmap_remove_all(), the mapping is not
  being destroyed, so the notion that the page was accessed is not lost.
  Moreover, clearing the page table entry's accessed bit and setting the
  page's PG_REFERENCED flag can throw off the page daemon's activity
  count calculation.  Finally, in my tests, I found that 15% of the
  atomic memory operations being performed by pmap_protect() were only
  to clear PG_A, and not change protection.  This could, by itself, be
  fixed, but I don't see the point given the above argument.
  
  Remove a comment from pmap_protect_pde() that is no longer meaningful
  after the above change.

Modified:
  head/sys/amd64/amd64/pmap.c
  head/sys/i386/i386/pmap.c

Modified: head/sys/amd64/amd64/pmap.c
==============================================================================
--- head/sys/amd64/amd64/pmap.c	Sun Apr 25 20:01:13 2010	(r207204)
+++ head/sys/amd64/amd64/pmap.c	Sun Apr 25 20:40:45 2010	(r207205)
@@ -2833,18 +2833,9 @@ retry:
 	if (oldpde & PG_MANAGED) {
 		eva = sva + NBPDR;
 		for (va = sva, m = PHYS_TO_VM_PAGE(oldpde & PG_PS_FRAME);
-		    va < eva; va += PAGE_SIZE, m++) {
-			/*
-			 * In contrast to the analogous operation on a 4KB page
-			 * mapping, the mapping's PG_A flag is not cleared and
-			 * the page's PG_REFERENCED flag is not set.  The
-			 * reason is that pmap_demote_pde() expects that a 2MB
-			 * page mapping with a stored page table page has PG_A
-			 * set.
-			 */
+		    va < eva; va += PAGE_SIZE, m++)
 			if ((oldpde & (PG_M | PG_RW)) == (PG_M | PG_RW))
 				vm_page_dirty(m);
-		}
 	}
 	if ((prot & VM_PROT_WRITE) == 0)
 		newpde &= ~(PG_RW | PG_M);
@@ -2953,23 +2944,15 @@ retry:
 			obits = pbits = *pte;
 			if ((pbits & PG_V) == 0)
 				continue;
-			if (pbits & PG_MANAGED) {
-				m = NULL;
-				if (pbits & PG_A) {
+
+			if ((prot & VM_PROT_WRITE) == 0) {
+				if ((pbits & (PG_MANAGED | PG_M | PG_RW)) ==
+				    (PG_MANAGED | PG_M | PG_RW)) {
 					m = PHYS_TO_VM_PAGE(pbits & PG_FRAME);
-					vm_page_flag_set(m, PG_REFERENCED);
-					pbits &= ~PG_A;
-				}
-				if ((pbits & (PG_M | PG_RW)) == (PG_M | PG_RW)) {
-					if (m == NULL)
-						m = PHYS_TO_VM_PAGE(pbits &
-						    PG_FRAME);
 					vm_page_dirty(m);
 				}
-			}
-
-			if ((prot & VM_PROT_WRITE) == 0)
 				pbits &= ~(PG_RW | PG_M);
+			}
 			if ((prot & VM_PROT_EXECUTE) == 0)
 				pbits |= pg_nx;
 

Modified: head/sys/i386/i386/pmap.c
==============================================================================
--- head/sys/i386/i386/pmap.c	Sun Apr 25 20:01:13 2010	(r207204)
+++ head/sys/i386/i386/pmap.c	Sun Apr 25 20:40:45 2010	(r207205)
@@ -2955,18 +2955,9 @@ retry:
 	if (oldpde & PG_MANAGED) {
 		eva = sva + NBPDR;
 		for (va = sva, m = PHYS_TO_VM_PAGE(oldpde & PG_PS_FRAME);
-		    va < eva; va += PAGE_SIZE, m++) {
-			/*
-			 * In contrast to the analogous operation on a 4KB page
-			 * mapping, the mapping's PG_A flag is not cleared and
-			 * the page's PG_REFERENCED flag is not set.  The
-			 * reason is that pmap_demote_pde() expects that a 2/4MB
-			 * page mapping with a stored page table page has PG_A
-			 * set.
-			 */
+		    va < eva; va += PAGE_SIZE, m++)
 			if ((oldpde & (PG_M | PG_RW)) == (PG_M | PG_RW))
 				vm_page_dirty(m);
-		}
 	}
 	if ((prot & VM_PROT_WRITE) == 0)
 		newpde &= ~(PG_RW | PG_M);
@@ -3074,22 +3065,15 @@ retry:
 			obits = pbits = *pte;
 			if ((pbits & PG_V) == 0)
 				continue;
-			if (pbits & PG_MANAGED) {
-				m = NULL;
-				if (pbits & PG_A) {
+
+			if ((prot & VM_PROT_WRITE) == 0) {
+				if ((pbits & (PG_MANAGED | PG_M | PG_RW)) ==
+				    (PG_MANAGED | PG_M | PG_RW)) {
 					m = PHYS_TO_VM_PAGE(pbits & PG_FRAME);
-					vm_page_flag_set(m, PG_REFERENCED);
-					pbits &= ~PG_A;
-				}
-				if ((pbits & (PG_M | PG_RW)) == (PG_M | PG_RW)) {
-					if (m == NULL)
-						m = PHYS_TO_VM_PAGE(pbits & PG_FRAME);
 					vm_page_dirty(m);
 				}
-			}
-
-			if ((prot & VM_PROT_WRITE) == 0)
 				pbits &= ~(PG_RW | PG_M);
+			}
 #ifdef PAE
 			if ((prot & VM_PROT_EXECUTE) == 0)
 				pbits |= pg_nx;

Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201004252040.o3PKejOe065064>