From owner-svn-src-head@FreeBSD.ORG Tue Jul 1 20:00:36 2014 Return-Path: Delivered-To: svn-src-head@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [8.8.178.115]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 87E4A53C; Tue, 1 Jul 2014 20:00:36 +0000 (UTC) Received: from svn.freebsd.org (svn.freebsd.org [IPv6:2001:1900:2254:2068::e6a:0]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mx1.freebsd.org (Postfix) with ESMTPS id 744302924; Tue, 1 Jul 2014 20:00:36 +0000 (UTC) Received: from svn.freebsd.org ([127.0.1.70]) by svn.freebsd.org (8.14.8/8.14.8) with ESMTP id s61K0aQ9067441; Tue, 1 Jul 2014 20:00:36 GMT (envelope-from bdrewery@svn.freebsd.org) Received: (from bdrewery@localhost) by svn.freebsd.org (8.14.8/8.14.8/Submit) id s61K0aCO067439; Tue, 1 Jul 2014 20:00:36 GMT (envelope-from bdrewery@svn.freebsd.org) Message-Id: <201407012000.s61K0aCO067439@svn.freebsd.org> From: Bryan Drewery Date: Tue, 1 Jul 2014 20:00:36 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r268114 - in head/sys: fs/nfsserver nfsserver X-SVN-Group: head MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-BeenThere: svn-src-head@freebsd.org X-Mailman-Version: 2.1.18 Precedence: list List-Id: SVN commit messages for the src tree for head/-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 01 Jul 2014 20:00:36 -0000 Author: bdrewery Date: Tue Jul 1 20:00:35 2014 New Revision: 268114 URL: http://svnweb.freebsd.org/changeset/base/268114 Log: Change NFS readdir() to only ignore cookies preceding the given offset for UFS rather than for all but ZFS. This code was assuming that offsets were monotonically increasing for all file systems except ZFS and that the cookies from a previous call may have been rewound to a block boundary. According to mckusick@ only UFS is known to do this, so only requests against UFS file systems should remove cookies smaller than the given offset. This fixes serving TMPFS over NFS as it too does not have monotonically increasing offsets. The comment around the code also indicated it was specific to UFS. Some of the code using 'not_zfs' is specific to ZFS snapshot handling, so add a 'is_zfs' variable for those cases. It's possible that 'is_zfs' check for VFS_VGET() support may not be specific to ZFS. This needs more research and testing. After this fix TMPFS and other file systems can be served over NFS. To test I compared the results of syncing a /usr/src tree into a tmpfs and serving that over NFS. Before the fix 3589 files were missing on the remote view. After the fix all files were successfully found. Reviewed by: rmacklem Discussed with: mckusick, rmacklem via fs@ Discussed at: http://lists.freebsd.org/pipermail/freebsd-fs/2014-April/019264.html MFC after: 2 weeks Sponsored by: EMC / Isilon Storage Division Modified: head/sys/fs/nfsserver/nfs_nfsdport.c head/sys/nfsserver/nfs_serv.c Modified: head/sys/fs/nfsserver/nfs_nfsdport.c ============================================================================== --- head/sys/fs/nfsserver/nfs_nfsdport.c Tue Jul 1 19:50:47 2014 (r268113) +++ head/sys/fs/nfsserver/nfs_nfsdport.c Tue Jul 1 20:00:35 2014 (r268114) @@ -1551,7 +1551,7 @@ nfsrvd_readdir(struct nfsrv_descript *nd u_long *cookies = NULL, *cookiep; struct uio io; struct iovec iv; - int not_zfs; + int is_ufs; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &at); @@ -1606,7 +1606,7 @@ nfsrvd_readdir(struct nfsrv_descript *nd nfsrv_postopattr(nd, getret, &at); goto out; } - not_zfs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "zfs"); + is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; MALLOC(rbuf, caddr_t, siz, M_TEMP, M_WAITOK); again: eofflag = 0; @@ -1686,12 +1686,10 @@ again: * skip over the records that precede the requested offset. This * requires the assumption that file offset cookies monotonically * increase. - * Since the offset cookies don't monotonically increase for ZFS, - * this is not done when ZFS is the file system. */ while (cpos < cend && ncookies > 0 && (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (not_zfs != 0 && ((u_quad_t)(*cookiep)) <= toff))) { + (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff))) { cpos += dp->d_reclen; dp = (struct dirent *)cpos; cookiep++; @@ -1804,7 +1802,7 @@ nfsrvd_readdirplus(struct nfsrv_descript struct uio io; struct iovec iv; struct componentname cn; - int at_root, needs_unbusy, not_zfs, supports_nfsv4acls; + int at_root, is_ufs, is_zfs, needs_unbusy, supports_nfsv4acls; struct mount *mp, *new_mp; uint64_t mounted_on_fileno; @@ -1884,7 +1882,8 @@ nfsrvd_readdirplus(struct nfsrv_descript nfsrv_postopattr(nd, getret, &at); goto out; } - not_zfs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "zfs"); + is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; + is_zfs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "zfs") == 0; MALLOC(rbuf, caddr_t, siz, M_TEMP, M_WAITOK); again: @@ -1957,12 +1956,10 @@ again: * skip over the records that precede the requested offset. This * requires the assumption that file offset cookies monotonically * increase. - * Since the offset cookies don't monotonically increase for ZFS, - * this is not done when ZFS is the file system. */ while (cpos < cend && ncookies > 0 && (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (not_zfs != 0 && ((u_quad_t)(*cookiep)) <= toff) || + (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff) || ((nd->nd_flag & ND_NFSV4) && ((dp->d_namlen == 1 && dp->d_name[0] == '.') || (dp->d_namlen==2 && dp->d_name[0]=='.' && dp->d_name[1]=='.'))))) { @@ -2004,7 +2001,7 @@ again: * This needs to be done here for NFSv4, since NFSv4 never does * a VFS_VGET() for "." or "..". */ - if (not_zfs == 0) { + if (is_zfs == 1) { r = VFS_VGET(mp, at.na_fileid, LK_SHARED, &nvp); if (r == EOPNOTSUPP) { usevget = 0; @@ -2153,7 +2150,7 @@ again: if (!r) r = nfsvno_getattr(nvp, nvap, nd->nd_cred, p, 1); - if (r == 0 && not_zfs == 0 && + if (r == 0 && is_zfs == 1 && nfsrv_enable_crossmntpt != 0 && (nd->nd_flag & ND_NFSV4) != 0 && nvp->v_type == VDIR && Modified: head/sys/nfsserver/nfs_serv.c ============================================================================== --- head/sys/nfsserver/nfs_serv.c Tue Jul 1 19:50:47 2014 (r268113) +++ head/sys/nfsserver/nfs_serv.c Tue Jul 1 20:00:35 2014 (r268114) @@ -2627,7 +2627,7 @@ nfsrv_readdir(struct nfsrv_descript *nfs int v3 = (nfsd->nd_flag & ND_NFSV3); u_quad_t off, toff, verf; u_long *cookies = NULL, *cookiep; /* needs to be int64_t or off_t */ - int not_zfs; + int is_ufs; nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); fhp = &nfh.fh_generic; @@ -2690,7 +2690,7 @@ nfsrv_readdir(struct nfsrv_descript *nfs error = 0; goto nfsmout; } - not_zfs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "zfs") != 0; + is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; VOP_UNLOCK(vp, 0); /* @@ -2777,12 +2777,10 @@ again: * skip over the records that precede the requested offset. This * requires the assumption that file offset cookies monotonically * increase. - * Since the offset cookies don't monotonically increase for ZFS, - * this is not done when ZFS is the file system. */ while (cpos < cend && ncookies > 0 && (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (not_zfs != 0 && ((u_quad_t)(*cookiep)) <= toff))) { + (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff))) { cpos += dp->d_reclen; dp = (struct dirent *)cpos; cookiep++; @@ -2928,7 +2926,7 @@ nfsrv_readdirplus(struct nfsrv_descript int usevget = 1; struct componentname cn; struct mount *mntp = NULL; - int not_zfs; + int is_ufs; nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); vp_locked = 0; @@ -2988,7 +2986,7 @@ nfsrv_readdirplus(struct nfsrv_descript error = 0; goto nfsmout; } - not_zfs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "zfs") != 0; + is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; VOP_UNLOCK(vp, 0); vp_locked = 0; rbuf = malloc(siz, M_TEMP, M_WAITOK); @@ -3068,12 +3066,10 @@ again: * skip over the records that precede the requested offset. This * requires the assumption that file offset cookies monotonically * increase. - * Since the offset cookies don't monotonically increase for ZFS, - * this is not done when ZFS is the file system. */ while (cpos < cend && ncookies > 0 && (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (not_zfs != 0 && ((u_quad_t)(*cookiep)) <= toff))) { + (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff))) { cpos += dp->d_reclen; dp = (struct dirent *)cpos; cookiep++;