[src/trunk]: src/sys/uvm In uvm_pagermapin(), pass VM_PROT_READ|VM_PROT_WRITE...

[thorpej <thorpej%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/fa84080046c7
branches:  trunk
changeset: 473240:fa84080046c7
user:      thorpej <thorpej%NetBSD.org@localhost>
date:      Wed May 26 06:42:57 1999 +0000

description:
In uvm_pagermapin(), pass VM_PROT_READ|VM_PROT_WRITE as access_type, to
ensure we don't take mod/ref emulation faults in an interrupt context
(e.g. during the i/o operation).  This is safe because:
        - For a pageout operation, the page is already known to be
          modified, and the pagedaemon will pmap_clear_modify() after
          the pageout has completed.
        - For a pagein operation, pagers must already pmap_clear_modify()
          after the pagein operation is complete, because the i/o may have
          been done with e.g. programmed i/o.
XXX It would be nice to know the i/o direction so that we can call
XXX pmap_enter() with only the protection and access_type necessary.

diffstat:

 sys/uvm/uvm_pager.c |  13 +++++++++++--
 1 files changed, 11 insertions(+), 2 deletions(-)

diffs (35 lines):

diff -r c804b682b9e6 -r fa84080046c7 sys/uvm/uvm_pager.c
--- a/sys/uvm/uvm_pager.c       Wed May 26 06:22:03 1999 +0000
+++ b/sys/uvm/uvm_pager.c       Wed May 26 06:42:57 1999 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: uvm_pager.c,v 1.18 1999/05/24 23:36:23 thorpej Exp $   */
+/*     $NetBSD: uvm_pager.c,v 1.19 1999/05/26 06:42:57 thorpej Exp $   */
 
 /*
  *
@@ -116,6 +116,9 @@
  *
  * we basically just map in a blank map entry to reserve the space in the
  * map and then use pmap_enter() to put the mappings in by hand.
+ *
+ * XXX It would be nice to know the direction of the I/O, so that we can
+ * XXX map only what is necessary.
  */
 
 vaddr_t
@@ -172,8 +175,14 @@
                        panic("uvm_pagermapin: page not busy");
 #endif
 
+               /*
+                * XXX VM_PROT_DEFAULT includes VM_PROT_EXEC; is that
+                * XXX really necessary?  It could lead to unnecessary
+                * XXX instruction cache flushes.
+                */
                pmap_enter(vm_map_pmap(pager_map), cva, VM_PAGE_TO_PHYS(pp),
-                   VM_PROT_DEFAULT, TRUE, 0);
+                   VM_PROT_DEFAULT, TRUE,
+                   VM_PROT_READ | VM_PROT_WRITE);
        }
 
        UVMHIST_LOG(maphist, "<- done (KVA=0x%x)", kva,0,0,0);