[src/trunk]: src/sys/rump/librump/rumpvfs Use proper locking before unbusying...

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

details:   https://anonhg.NetBSD.org/src/rev/9e740fa5ea02
branches:  trunk
changeset: 757643:9e740fa5ea02
user:      pooka <pooka%NetBSD.org@localhost>
date:      Thu Sep 09 09:50:21 2010 +0000

description:
Use proper locking before unbusying pages.

Caught after yesterday's changes by the test suite (the ffs snapshot
test, to be precise).

diffstat:

 sys/rump/librump/rumpvfs/vm_vfs.c |  15 +++++++++++++--
 1 files changed, 13 insertions(+), 2 deletions(-)

diffs (49 lines):

diff -r 9e7021d7920a -r 9e740fa5ea02 sys/rump/librump/rumpvfs/vm_vfs.c
--- a/sys/rump/librump/rumpvfs/vm_vfs.c Thu Sep 09 09:12:35 2010 +0000
+++ b/sys/rump/librump/rumpvfs/vm_vfs.c Thu Sep 09 09:50:21 2010 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: vm_vfs.c,v 1.20 2010/09/06 21:33:07 pooka Exp $        */
+/*     $NetBSD: vm_vfs.c,v 1.21 2010/09/09 09:50:21 pooka Exp $        */
 
 /*
  * Copyright (c) 2008 Antti Kantee.  All Rights Reserved.
@@ -29,7 +29,7 @@
  */
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: vm_vfs.c,v 1.20 2010/09/06 21:33:07 pooka Exp $");
+__KERNEL_RCSID(0, "$NetBSD: vm_vfs.c,v 1.21 2010/09/09 09:50:21 pooka Exp $");
 
 #include <sys/param.h>
 
@@ -47,11 +47,13 @@
 void
 uvm_aio_aiodone(struct buf *bp)
 {
+       struct uvm_object *uobj;
        int i, npages = bp->b_bufsize >> PAGE_SHIFT;
        struct vm_page **pgs;
        vaddr_t va;
        int pageout = 0;
 
+       KASSERT(npages > 0);
        pgs = kmem_alloc(npages * sizeof(*pgs), KM_SLEEP);
        for (i = 0; i < npages; i++) {
                va = (vaddr_t)bp->b_data + (i << PAGE_SHIFT);
@@ -65,7 +67,16 @@
 
        uvm_pagermapout((vaddr_t)bp->b_data, npages);
        uvm_pageout_done(pageout);
+
+       /* get uobj because we need it after pages might be recycled */
+       uobj = pgs[0]->uobject;
+       KASSERT(uobj);
+
+       mutex_enter(&uobj->vmobjlock);
+       mutex_enter(&uvm_pageqlock);
        uvm_page_unbusy(pgs, npages);
+       mutex_exit(&uvm_pageqlock);
+       mutex_exit(&uobj->vmobjlock);
 
        if (BUF_ISWRITE(bp) && (bp->b_cflags & BC_AGE) != 0) {
                mutex_enter(bp->b_objlock);