From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-8.3 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS, USER_AGENT_SANE_1 autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 06E65C32751 for ; Wed, 31 Jul 2019 17:45:36 +0000 (UTC) Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id CFFFA2067D for ; Wed, 31 Jul 2019 17:45:35 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org CFFFA2067D Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Received: from localhost ([::1]:43304 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.86_2) (envelope-from ) id 1hssfO-0006qK-RC for qemu-devel@archiver.kernel.org; Wed, 31 Jul 2019 13:45:34 -0400 Received: from eggs.gnu.org ([2001:470:142:3::10]:39006) by lists.gnu.org with esmtp (Exim 4.86_2) (envelope-from ) id 1hsset-0006RL-93 for qemu-devel@nongnu.org; Wed, 31 Jul 2019 13:45:04 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1hsses-0003V9-3S for qemu-devel@nongnu.org; Wed, 31 Jul 2019 13:45:03 -0400 Received: from mx1.redhat.com ([209.132.183.28]:40306) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1hsser-0003Uh-SD for qemu-devel@nongnu.org; Wed, 31 Jul 2019 13:45:02 -0400 Received: from smtp.corp.redhat.com (int-mx08.intmail.prod.int.phx2.redhat.com [10.5.11.23]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id AB79330A7C6D for ; Wed, 31 Jul 2019 17:45:00 +0000 (UTC) Received: from work-vm (ovpn-116-252.ams2.redhat.com [10.36.116.252]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 8E20119C65; Wed, 31 Jul 2019 17:44:55 +0000 (UTC) Date: Wed, 31 Jul 2019 18:44:52 +0100 From: "Dr. David Alan Gilbert" To: Stefan Hajnoczi Message-ID: <20190731174452.GL3203@work-vm> References: <20190726091103.23503-1-stefanha@redhat.com> <20190726091103.23503-6-stefanha@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20190726091103.23503-6-stefanha@redhat.com> User-Agent: Mutt/1.12.0 (2019-05-25) X-Scanned-By: MIMEDefang 2.84 on 10.5.11.23 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.42]); Wed, 31 Jul 2019 17:45:00 +0000 (UTC) X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] X-Received-From: 209.132.183.28 Subject: Re: [Qemu-devel] [PATCH 5/5] virtiofsd: prevent races with lo_dirp_put() X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: virtio-fs@redhat.com, qemu-devel@nongnu.org Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: "Qemu-devel" * Stefan Hajnoczi (stefanha@redhat.com) wrote: > Introduce lo_dirp_put() so that FUSE_RELEASEDIR does not cause > use-after-free races with other threads that are accessing lo_dirp. > > Also make lo_releasedir() atomic to prevent FUSE_RELEASEDIR racing with > itself. This prevents double-frees. > > Signed-off-by: Stefan Hajnoczi > --- > contrib/virtiofsd/passthrough_ll.c | 42 +++++++++++++++++++++++++----- > 1 file changed, 36 insertions(+), 6 deletions(-) > > diff --git a/contrib/virtiofsd/passthrough_ll.c b/contrib/virtiofsd/passthrough_ll.c > index ad3abdd532..f74e7d2d21 100644 > --- a/contrib/virtiofsd/passthrough_ll.c > +++ b/contrib/virtiofsd/passthrough_ll.c > @@ -1293,11 +1293,28 @@ static void lo_readlink(fuse_req_t req, fuse_ino_t ino) > } > > struct lo_dirp { > + gint refcount; > DIR *dp; > struct dirent *entry; > off_t offset; > }; > > +static void lo_dirp_put(struct lo_dirp **dp) > +{ > + struct lo_dirp *d = *dp; > + > + if (!d) { > + return; > + } > + *dp = NULL; > + > + if (g_atomic_int_dec_and_test(&d->refcount)) { > + closedir(d->dp); > + free(d); > + } > +} > + > +/* Call lo_dirp_put() on the return value when no longer needed */ > static struct lo_dirp *lo_dirp(fuse_req_t req, struct fuse_file_info *fi) > { > struct lo_data *lo = lo_data(req); > @@ -1305,6 +1322,9 @@ static struct lo_dirp *lo_dirp(fuse_req_t req, struct fuse_file_info *fi) > > pthread_mutex_lock(&lo->mutex); > elem = lo_map_get(&lo->dirp_map, fi->fh); > + if (elem) { > + g_atomic_int_inc(&elem->dirp->refcount); I don't understand what protects against reading the elem->dirp here at the same time it's free'd by lo_releasedir's call to lo_dirp_put > + } > pthread_mutex_unlock(&lo->mutex); > if (!elem) > return NULL; > @@ -1335,6 +1355,8 @@ static void lo_opendir(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi > d->offset = 0; > d->entry = NULL; > > + g_atomic_int_set(&d->refcount, 1); /* paired with lo_releasedir() */ > + > fh = lo_add_dirp_mapping(req, d); > if (fh == -1) > goto out_err; > @@ -1363,7 +1385,7 @@ static void lo_do_readdir(fuse_req_t req, fuse_ino_t ino, size_t size, > off_t offset, struct fuse_file_info *fi, int plus) > { > struct lo_data *lo = lo_data(req); > - struct lo_dirp *d; > + struct lo_dirp *d = NULL; > struct lo_inode *dinode; > char *buf = NULL; > char *p; > @@ -1451,6 +1473,8 @@ static void lo_do_readdir(fuse_req_t req, fuse_ino_t ino, size_t size, > > err = 0; > error: > + lo_dirp_put(&d); > + > // If there's an error, we can only signal it if we haven't stored > // any entries yet - otherwise we'd end up with wrong lookup > // counts for the entries that are already in the buffer. So we > @@ -1477,22 +1501,25 @@ static void lo_readdirplus(fuse_req_t req, fuse_ino_t ino, size_t size, > static void lo_releasedir(fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi) > { > struct lo_data *lo = lo_data(req); > + struct lo_map_elem *elem; > struct lo_dirp *d; > > (void) ino; > > - d = lo_dirp(req, fi); > - if (!d) { > + pthread_mutex_lock(&lo->mutex); > + elem = lo_map_get(&lo->dirp_map, fi->fh); > + if (!elem) { > + pthread_mutex_unlock(&lo->mutex); > fuse_reply_err(req, EBADF); > return; > } > > - pthread_mutex_lock(&lo->mutex); > + d = elem->dirp; > lo_map_remove(&lo->dirp_map, fi->fh); > pthread_mutex_unlock(&lo->mutex); > > - closedir(d->dp); > - free(d); > + lo_dirp_put(&d); /* paired with lo_opendir() */ Is the &d really what's intended? That's the local stack variable, so lo_dirp_put will set that local to NULL rather than the elem->dirp wont it? Dave > + > fuse_reply_err(req, 0); > } > > @@ -1701,6 +1728,9 @@ static void lo_fsyncdir(fuse_req_t req, fuse_ino_t ino, int datasync, > res = fdatasync(fd); > else > res = fsync(fd); > + > + lo_dirp_put(&d); > + > fuse_reply_err(req, res == -1 ? errno : 0); > } > > -- > 2.21.0 > -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK