On Friday, August 09, 2013 4:40:10 pm Alan Cox wrote:
>
> On Aug 9, 2013, at 1:34 PM, Alan Cox wrote:
>
> >
> > On Aug 9, 2013, at 12:56 PM, John Baldwin wrote:
> >
> >> On Friday, August 09, 2013 12:43:50 pm David E. O'Brien wrote:
> >>> Author: obrien
> >>> Date: Fri Aug 9 16:43:50 2013
> >>> New Revision: 254150
> >>> URL: http://svnweb.freebsd.org/changeset/base/254150
> >>>
> >>> Log:
> >>> Add missing 'VPO_BUSY' from r254141 to fix kernel build break.
> >>>
> >>> Modified:
> >>> head/sys/vm/vm_page.h
> >>
> >> This can't possibly be correct as r254138 just removed this flag. If it
isn't
> >> obvious how to fix the uses added back in r254141, then r254141 should be
> >> reverted instead.
> >>
> >> Hmm, looking at the relevant bits of r254141, it doesn't look obvious:
> >>
> >> + /* Detach the old page from the resident tailq. */
> >> + TAILQ_REMOVE(&object->memq, mold, listq);
> >> + vm_page_lock(mold);
> >
> > Replace the next four lines with
> >
> > vm_page_xunbusy(mold);
> >
>
> On second thought, no, because it could lead to lock recursion.
What about this. I think this matches the common idiom I've seen in
other places.
Index: vm_page.c
===================================================================
--- vm_page.c (revision 254158)
+++ vm_page.c (working copy)
@@ -1202,12 +1202,9 @@
/* Detach the old page from the resident tailq. */
TAILQ_REMOVE(&object->memq, mold, listq);
vm_page_lock(mold);
- if (mold->oflags & VPO_BUSY) {
- mold->oflags &= ~VPO_BUSY;
- vm_page_flash(mold);
- }
mold->object = NULL;
vm_page_unlock(mold);
+ vm_page_xunbusy(mold);
/* Insert the new page in the resident tailq. */
if (mpred != NULL)
--
John Baldwin
_______________________________________________
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
