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=-6.9 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH, MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS 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 6E4D5C00454 for ; Thu, 12 Dec 2019 17:13:32 +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 3F8992253D for ; Thu, 12 Dec 2019 17:13:32 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="Nt9GFHwU" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 3F8992253D 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]:34496 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1ifS1q-0000op-OL for qemu-devel@archiver.kernel.org; Thu, 12 Dec 2019 12:13:30 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]:50142) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1ifRVg-0005pv-GJ for qemu-devel@nongnu.org; Thu, 12 Dec 2019 11:40:18 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ifRVe-0002CJ-L2 for qemu-devel@nongnu.org; Thu, 12 Dec 2019 11:40:16 -0500 Received: from us-smtp-delivery-1.mimecast.com ([207.211.31.120]:57737 helo=us-smtp-1.mimecast.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1ifRVc-00029B-Hg for qemu-devel@nongnu.org; Thu, 12 Dec 2019 11:40:13 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1576168811; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=KjTJlOeSDvTMFG+Z7IYeIYkmtBqqT9LFbg4f4QaR7T4=; b=Nt9GFHwU+T7akoL4M4Pg9eqfp0jaCn+DI/LQjN7a5DmM1ehjHHZnOdRRCyCMRtZIeUV9Md XZgg6EwMB/0eV9CtB1ovCeJZS7Z6YspJdL+QlBZCXbCEK+eEWUT7FlDfXAJgxObtPDgW+L 0OhQQDtmbzvtOA/AZcr0C5hd4pyeM1M= Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-283-kHbCYqpqNO63Ntm8fJAO4Q-1; Thu, 12 Dec 2019 11:40:10 -0500 Received: from smtp.corp.redhat.com (int-mx02.intmail.prod.int.phx2.redhat.com [10.5.11.12]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id D51FE477 for ; Thu, 12 Dec 2019 16:40:09 +0000 (UTC) Received: from dgilbert-t580.localhost (ovpn-116-226.ams2.redhat.com [10.36.116.226]) by smtp.corp.redhat.com (Postfix) with ESMTP id 0107860BF3; Thu, 12 Dec 2019 16:40:08 +0000 (UTC) From: "Dr. David Alan Gilbert (git)" To: qemu-devel@nongnu.org, stefanha@redhat.com, vgoyal@redhat.com Subject: [PATCH 037/104] virtiofsd: passthrough_ll: add fallback for racy ops Date: Thu, 12 Dec 2019 16:37:57 +0000 Message-Id: <20191212163904.159893-38-dgilbert@redhat.com> In-Reply-To: <20191212163904.159893-1-dgilbert@redhat.com> References: <20191212163904.159893-1-dgilbert@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.79 on 10.5.11.12 X-MC-Unique: kHbCYqpqNO63Ntm8fJAO4Q-1 X-Mimecast-Spam-Score: 0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] [fuzzy] X-Received-From: 207.211.31.120 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: , Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: "Qemu-devel" From: Miklos Szeredi We have two operations that cannot be done race-free on a symlink in certain cases: utimes and link. Add racy fallback for these if the race-free method doesn't work. We do our best to avoid races even in this case: - get absolute path by reading /proc/self/fd/NN symlink - lookup parent directory: after this we are safe against renames in ancestors - lookup name in parent directory, and verify that we got to the original inode, if not retry the whole thing Both utimes(2) and link(2) hold i_lock on the inode across the operation, so a racing rename/delete by this fuse instance is not possible, only from other entities changing the filesystem. If the "norace" option is given, then disable the racy fallbacks. Signed-off-by: Miklos Szeredi --- tools/virtiofsd/passthrough_ll.c | 159 +++++++++++++++++++++++++++---- 1 file changed, 142 insertions(+), 17 deletions(-) diff --git a/tools/virtiofsd/passthrough_ll.c b/tools/virtiofsd/passthrough= _ll.c index 93e74cce21..1faae2753f 100644 --- a/tools/virtiofsd/passthrough_ll.c +++ b/tools/virtiofsd/passthrough_ll.c @@ -98,6 +98,7 @@ enum { struct lo_data { pthread_mutex_t mutex; int debug; + int norace; int writeback; int flock; int xattr; @@ -124,10 +125,15 @@ static const struct fuse_opt lo_opts[] =3D { { "cache=3Dnever", offsetof(struct lo_data, cache), CACHE_NEVER }, { "cache=3Dauto", offsetof(struct lo_data, cache), CACHE_NORMAL }, { "cache=3Dalways", offsetof(struct lo_data, cache), CACHE_ALWAYS }, - + { "norace", offsetof(struct lo_data, norace), 1 }, FUSE_OPT_END }; =20 +static void unref_inode(struct lo_data *lo, struct lo_inode *inode, uint64= _t n); + +static struct lo_inode *lo_find(struct lo_data *lo, struct stat *st); + + static struct lo_data *lo_data(fuse_req_t req) { return (struct lo_data *)fuse_req_userdata(req); @@ -347,23 +353,127 @@ static void lo_getattr(fuse_req_t req, fuse_ino_t in= o, fuse_reply_attr(req, &buf, lo->timeout); } =20 -static int utimensat_empty_nofollow(struct lo_inode *inode, - const struct timespec *tv) +static int lo_parent_and_name(struct lo_data *lo, struct lo_inode *inode, + char path[PATH_MAX], struct lo_inode **paren= t) { - int res; char procname[64]; + char *last; + struct stat stat; + struct lo_inode *p; + int retries =3D 2; + int res; + +retry: + sprintf(procname, "/proc/self/fd/%i", inode->fd); + + res =3D readlink(procname, path, PATH_MAX); + if (res < 0) { + fuse_log(FUSE_LOG_WARNING, "lo_parent_and_name: readlink failed: %= m\n"); + goto fail_noretry; + } + + if (res >=3D PATH_MAX) { + fuse_log(FUSE_LOG_WARNING, "lo_parent_and_name: readlink overflowe= d\n"); + goto fail_noretry; + } + path[res] =3D '\0'; + + last =3D strrchr(path, '/'); + if (last =3D=3D NULL) { + /* Shouldn't happen */ + fuse_log( + FUSE_LOG_WARNING, + "lo_parent_and_name: INTERNAL ERROR: bad path read from proc\n= "); + goto fail_noretry; + } + if (last =3D=3D path) { + p =3D &lo->root; + pthread_mutex_lock(&lo->mutex); + p->refcount++; + pthread_mutex_unlock(&lo->mutex); + } else { + *last =3D '\0'; + res =3D fstatat(AT_FDCWD, last =3D=3D path ? "/" : path, &stat, 0)= ; + if (res =3D=3D -1) { + if (!retries) { + fuse_log(FUSE_LOG_WARNING, + "lo_parent_and_name: failed to stat parent: %m\n"= ); + } + goto fail; + } + p =3D lo_find(lo, &stat); + if (p =3D=3D NULL) { + if (!retries) { + fuse_log(FUSE_LOG_WARNING, + "lo_parent_and_name: failed to find parent\n"); + } + goto fail; + } + } + last++; + res =3D fstatat(p->fd, last, &stat, AT_SYMLINK_NOFOLLOW); + if (res =3D=3D -1) { + if (!retries) { + fuse_log(FUSE_LOG_WARNING, + "lo_parent_and_name: failed to stat last\n"); + } + goto fail_unref; + } + if (stat.st_dev !=3D inode->dev || stat.st_ino !=3D inode->ino) { + if (!retries) { + fuse_log(FUSE_LOG_WARNING, + "lo_parent_and_name: failed to match last\n"); + } + goto fail_unref; + } + *parent =3D p; + memmove(path, last, strlen(last) + 1); + + return 0; + +fail_unref: + unref_inode(lo, p, 1); +fail: + if (retries) { + retries--; + goto retry; + } +fail_noretry: + errno =3D EIO; + return -1; +} + +static int utimensat_empty(struct lo_data *lo, struct lo_inode *inode, + const struct timespec *tv) +{ + int res; + struct lo_inode *parent; + char path[PATH_MAX]; =20 if (inode->is_symlink) { - res =3D utimensat(inode->fd, "", tv, AT_EMPTY_PATH | AT_SYMLINK_NO= FOLLOW); + res =3D utimensat(inode->fd, "", tv, AT_EMPTY_PATH); if (res =3D=3D -1 && errno =3D=3D EINVAL) { /* Sorry, no race free way to set times on symlink. */ - errno =3D EPERM; + if (lo->norace) { + errno =3D EPERM; + } else { + goto fallback; + } } return res; } - sprintf(procname, "/proc/self/fd/%i", inode->fd); + sprintf(path, "/proc/self/fd/%i", inode->fd); + + return utimensat(AT_FDCWD, path, tv, 0); =20 - return utimensat(AT_FDCWD, procname, tv, 0); +fallback: + res =3D lo_parent_and_name(lo, inode, path, &parent); + if (res !=3D -1) { + res =3D utimensat(parent->fd, path, tv, AT_SYMLINK_NOFOLLOW); + unref_inode(lo, parent, 1); + } + + return res; } =20 static int lo_fi_fd(fuse_req_t req, struct fuse_file_info *fi) @@ -387,6 +497,7 @@ static void lo_setattr(fuse_req_t req, fuse_ino_t ino, = struct stat *attr, { int saverr; char procname[64]; + struct lo_data *lo =3D lo_data(req); struct lo_inode *inode; int ifd; int res; @@ -459,7 +570,7 @@ static void lo_setattr(fuse_req_t req, fuse_ino_t ino, = struct stat *attr, if (fi) { res =3D futimens(fd, tv); } else { - res =3D utimensat_empty_nofollow(inode, tv); + res =3D utimensat_empty(lo, inode, tv); } if (res =3D=3D -1) { goto out_err; @@ -692,24 +803,38 @@ static void lo_symlink(fuse_req_t req, const char *li= nk, fuse_ino_t parent, lo_mknod_symlink(req, parent, name, S_IFLNK, 0, link); } =20 -static int linkat_empty_nofollow(struct lo_inode *inode, int dfd, - const char *name) +static int linkat_empty_nofollow(struct lo_data *lo, struct lo_inode *inod= e, + int dfd, const char *name) { int res; - char procname[64]; + struct lo_inode *parent; + char path[PATH_MAX]; =20 if (inode->is_symlink) { res =3D linkat(inode->fd, "", dfd, name, AT_EMPTY_PATH); if (res =3D=3D -1 && (errno =3D=3D ENOENT || errno =3D=3D EINVAL))= { /* Sorry, no race free way to hard-link a symlink. */ - errno =3D EPERM; + if (lo->norace) { + errno =3D EPERM; + } else { + goto fallback; + } } return res; } =20 - sprintf(procname, "/proc/self/fd/%i", inode->fd); + sprintf(path, "/proc/self/fd/%i", inode->fd); + + return linkat(AT_FDCWD, path, dfd, name, AT_SYMLINK_FOLLOW); + +fallback: + res =3D lo_parent_and_name(lo, inode, path, &parent); + if (res !=3D -1) { + res =3D linkat(parent->fd, path, dfd, name, 0); + unref_inode(lo, parent, 1); + } =20 - return linkat(AT_FDCWD, procname, dfd, name, AT_SYMLINK_FOLLOW); + return res; } =20 static void lo_link(fuse_req_t req, fuse_ino_t ino, fuse_ino_t parent, @@ -731,7 +856,7 @@ static void lo_link(fuse_req_t req, fuse_ino_t ino, fus= e_ino_t parent, e.attr_timeout =3D lo->timeout; e.entry_timeout =3D lo->timeout; =20 - res =3D linkat_empty_nofollow(inode, lo_fd(req, parent), name); + res =3D linkat_empty_nofollow(lo, inode, lo_fd(req, parent), name); if (res =3D=3D -1) { goto out_err; } @@ -1544,7 +1669,7 @@ static void lo_setxattr(fuse_req_t req, fuse_ino_t in= o, const char *name, } =20 if (inode->is_symlink) { - /* Sorry, no race free way to setxattr on symlink. */ + /* Sorry, no race free way to removexattr on symlink. */ saverr =3D EPERM; goto out; } --=20 2.23.0