From owner-svn-src-all@FreeBSD.ORG Mon Jun 18 22:17:29 2012 Return-Path: Delivered-To: svn-src-all@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id BBB7C1065675; Mon, 18 Jun 2012 22:17:29 +0000 (UTC) (envelope-from rmacklem@FreeBSD.org) Received: from svn.freebsd.org (svn.freebsd.org [IPv6:2001:4f8:fff6::2c]) by mx1.freebsd.org (Postfix) with ESMTP id 8DE038FC0A; Mon, 18 Jun 2012 22:17:29 +0000 (UTC) Received: from svn.freebsd.org (localhost [127.0.0.1]) by svn.freebsd.org (8.14.4/8.14.4) with ESMTP id q5IMHTaj003930; Mon, 18 Jun 2012 22:17:29 GMT (envelope-from rmacklem@svn.freebsd.org) Received: (from rmacklem@localhost) by svn.freebsd.org (8.14.4/8.14.4/Submit) id q5IMHTJE003928; Mon, 18 Jun 2012 22:17:29 GMT (envelope-from rmacklem@svn.freebsd.org) Message-Id: <201206182217.q5IMHTJE003928@svn.freebsd.org> From: Rick Macklem Date: Mon, 18 Jun 2012 22:17:29 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org X-SVN-Group: head MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: Subject: svn commit: r237244 - head/sys/fs/nfsclient X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 18 Jun 2012 22:17:29 -0000 Author: rmacklem Date: Mon Jun 18 22:17:28 2012 New Revision: 237244 URL: http://svn.freebsd.org/changeset/base/237244 Log: Fix the NFSv4 client for the case where mmap'd files are written, but not msync'd by a process. A VOP_PUTPAGES() called when VOP_RECLAIM() happens will usually fail, since the NFSv4 Open has already been closed by VOP_INACTIVE(). Add a vm_object_page_clean() call to the NFSv4 client's VOP_INACTIVE(), so that the write happens before the NFSv4 Open is closed. kib@ suggested using vgone() instead and I will explore this, but this patch fixes things in the meantime. For some reason, the VOP_PUTPAGES() is still attaempted in VOP_RECLAIM(), but having this fail doesn't cause any problems except a "stateid0 in write" being logged. Reviewed by: kib MFC after: 1 week Modified: head/sys/fs/nfsclient/nfs_clnode.c Modified: head/sys/fs/nfsclient/nfs_clnode.c ============================================================================== --- head/sys/fs/nfsclient/nfs_clnode.c Mon Jun 18 21:08:48 2012 (r237243) +++ head/sys/fs/nfsclient/nfs_clnode.c Mon Jun 18 22:17:28 2012 (r237244) @@ -210,18 +210,28 @@ ncl_inactive(struct vop_inactive_args *a struct nfsnode *np; struct sillyrename *sp; struct vnode *vp = ap->a_vp; + boolean_t retv; np = VTONFS(vp); if (NFS_ISV4(vp) && vp->v_type == VREG) { /* * Since mmap()'d files do I/O after VOP_CLOSE(), the NFSv4 - * Close operations are delayed until now. Any dirty buffers - * must be flushed before the close, so that the stateid is - * available for the writes. + * Close operations are delayed until now. Any dirty + * buffers/pages must be flushed before the close, so that the + * stateid is available for the writes. */ - (void) ncl_flush(vp, MNT_WAIT, NULL, ap->a_td, 1, 0); - (void) nfsrpc_close(vp, 1, ap->a_td); + if (vp->v_object != NULL) { + VM_OBJECT_LOCK(vp->v_object); + retv = vm_object_page_clean(vp->v_object, 0, 0, + OBJPC_SYNC); + VM_OBJECT_UNLOCK(vp->v_object); + } else + retv = TRUE; + if (retv == TRUE) { + (void)ncl_flush(vp, MNT_WAIT, NULL, ap->a_td, 1, 0); + (void)nfsrpc_close(vp, 1, ap->a_td); + } } mtx_lock(&np->n_mtx);