Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Chuck Lever III <chuck.lever@oracle.com>
To:     Neil Brown <neilb@suse.de>
CC:     Petr Vorel <pvorel@suse.cz>, Bruce Fields <bfields@fieldses.org>,
        Linux NFS Mailing List <linux-nfs@vger.kernel.org>,
        Steve Dickson <steved@redhat.com>,
        Alexey Kodanev <alexey.kodanev@oracle.com>
Subject: Re: [PATCH/RFC nfs-utils] Fix NFSv4 export of tmpfs filesystems.
Thread-Topic: [PATCH/RFC nfs-utils] Fix NFSv4 export of tmpfs filesystems.
Thread-Index: AQHXQuMiXfq5XDoV1UyaLfWrKHICQKrYC2+A
Date:   Fri, 7 May 2021 13:55:16 +0000
Message-ID: <A4E16B80-F754-4F11-9689-C2A4E5588019@oracle.com>
References: <20210422191803.31511-1-pvorel@suse.cz>
 <20210422202334.GB25415@fieldses.org> <YILQip3nAxhpXP9+@pevik>
 <162035212343.24322.12361160756597283121@noble.neil.brown.name>
In-Reply-To: <162035212343.24322.12361160756597283121@noble.neil.brown.name>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-ID: <28C26E98364F924D944EEAA5CA24627D@namprd10.prod.outlook.com>
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 4c8776a7-6987-4312-ffca-08d9115fbe5f
X-MS-Exchange-CrossTenant-originalarrivaltime: 07 May 2021 13:55:16.2018
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 4e2c6054-71cb-48f1-bd6c-3a9705aca71b
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: cUt/J5Rg5ui/sVvASDGzmzyRZQcnAoypMLHArvsKv7tVEhpwUYom8270WknP9N1/qT0U2091z9f+W3gRkrf2qg==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SJ0PR10MB5485
X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=9977 signatures=668683
X-Proofpoint-Spam-Details: rule=notspam policy=default score=0 bulkscore=0 adultscore=0 phishscore=0
 spamscore=0 suspectscore=0 mlxscore=0 mlxlogscore=999 malwarescore=0
 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2104190000
 definitions=main-2105070095
X-Proofpoint-GUID: 5smi41rr_WozuWak2Exaha9HmA18GW9Y
X-Proofpoint-ORIG-GUID: 5smi41rr_WozuWak2Exaha9HmA18GW9Y
X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=9977 signatures=668683
X-Proofpoint-Spam-Details: rule=notspam policy=default score=0 malwarescore=0 mlxscore=0 adultscore=0
 clxscore=1011 spamscore=0 mlxlogscore=999 priorityscore=1501
 lowpriorityscore=0 phishscore=0 suspectscore=0 impostorscore=0 bulkscore=0
 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2104190000
 definitions=main-2105070095
Precedence: bulk
List-ID: <linux-nfs.vger.kernel.org>
X-Mailing-List: linux-nfs@vger.kernel.org


> On May 6, 2021, at 9:48 PM, NeilBrown <neilb@suse.de> wrote:
>=20
>=20
> [[This is a proposed fix.  It seems to work.  I'd like
>  some review comments before it is committed.
>  Petr: it would be great if you could test it to confirm
>  it actually works in your case.
> ]]
>=20
> Some filesystems cannot be exported without an fsid or uuid.
> tmpfs is the main example.
>=20
> When mountd creates nfsv4 pseudo-root exports for the path leading down
> to an export point it exports each directory without any fsid or uuid.
> If one of these directories is on tmp, that will fail.
>=20
> The net result is that exporting a subdirectory of a tmpfs filesystem
> will not work over NFSv4 as the parents within the filesystem cannot be
> exported.  It will either fail, or fall-back to NFSv3 (depending on the
> version of the mount.nfs program).
>=20
> To fix this we need to provide an fsid or uuid for these pseudo-root
> exports.  This patch does that by creating a UUID with the first 4 bytes
> 0xFFFFFFFF and the remaining 12 bytes form from the path name, xoring
> bytes together if the path is longer than 12 characters.
> Hopefully no filesystem uses a UUID like this....

That's not really a UUID, as per RFC 4122. I'm guessing it's possible
for a collision to occur pretty quickly, for instance. It would be nicer
if a conformant UUID could be used here.

Is there a problem with specifying the export's fsid in /etc/exports?


> The patch borrows some code from exportfs.  Maybe that code should be
> move to a library..
>=20
> Signed-off-by: NeilBrown <neilb@suse.de>
> ---
> support/export/v4root.c | 57 +++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 57 insertions(+)
>=20
> diff --git a/support/export/v4root.c b/support/export/v4root.c
> index 3654bd7c10c0..fd36eb704441 100644
> --- a/support/export/v4root.c
> +++ b/support/export/v4root.c
> @@ -11,6 +11,7 @@
> #include <config.h>
> #endif
>=20
> +#include <fcntl.h>
> #include <sys/types.h>
> #include <sys/stat.h>
> #include <sys/queue.h>
> @@ -21,6 +22,7 @@
> #include <unistd.h>
> #include <errno.h>
>=20
> +#include "nfsd_path.h"
> #include "xlog.h"
> #include "exportfs.h"
> #include "nfslib.h"
> @@ -73,6 +75,38 @@ set_pseudofs_security(struct exportent *pseudo)
> 	}
> }
>=20
> +static ssize_t exportfs_write(int fd, const char *buf, size_t len)
> +{
> +	return nfsd_path_write(fd, buf, len);
> +}
> +
> +static int test_export(struct exportent *eep, int with_fsid)
> +{
> +	char *path =3D eep->e_path;
> +	int flags =3D eep->e_flags | (with_fsid ? NFSEXP_FSID : 0);
> +	/* beside max path, buf size should take protocol str into account */
> +	char buf[NFS_MAXPATHLEN+1+64] =3D { 0 };
> +	char *bp =3D buf;
> +	int len =3D sizeof(buf);
> +	int fd, n;
> +
> +	n =3D snprintf(buf, len, "-test-client- ");
> +	bp +=3D n;
> +	len -=3D n;
> +	qword_add(&bp, &len, path);
> +	if (len < 1)
> +		return 0;
> +	snprintf(bp, len, " 3 %d 65534 65534 0\n", flags);
> +	fd =3D open("/proc/net/rpc/nfsd.export/channel", O_WRONLY);
> +	if (fd < 0)
> +		return 0;
> +	n =3D exportfs_write(fd, buf, strlen(buf));
> +	close(fd);
> +	if (n < 0)
> +		return 0;
> +	return 1;
> +}
> +
> /*
>  * Create a pseudo export
>  */
> @@ -82,6 +116,7 @@ v4root_create(char *path, nfs_export *export)
> 	nfs_export *exp;
> 	struct exportent eep;
> 	struct exportent *curexp =3D &export->m_export;
> +	char uuid[33];
>=20
> 	dupexportent(&eep, &pseudo_root.m_export);
> 	eep.e_ttl =3D default_ttl;
> @@ -89,6 +124,28 @@ v4root_create(char *path, nfs_export *export)
> 	strncpy(eep.e_path, path, sizeof(eep.e_path)-1);
> 	if (strcmp(path, "/") !=3D 0)
> 		eep.e_flags &=3D ~NFSEXP_FSID;
> +	if (strcmp(path, "/") !=3D 0 &&
> +	    !test_export(&eep, 0)) {
> +		/* Need a uuid - base it on path */
> +		char buf[12], *pp =3D path;
> +		unsigned int i =3D 0;
> +
> +		memset(buf, 0, sizeof(buf));
> +		while (*pp) {
> +			buf[i] ^=3D *pp++;
> +			i +=3D 1;
> +			if (i >=3D sizeof(buf))
> +				i =3D 0;
> +		}
> +		memset(uuid, 'F', 32);
> +		uuid[32] =3D '\0';
> +		pp =3D uuid + 32 - sizeof(buf) * 2;
> +		for (i =3D 0; i < sizeof(buf); i++) {
> +			snprintf(pp, 3, "%02X", buf[i]);
> +			pp +=3D 2;
> +		}
> +		eep.e_uuid =3D uuid;
> +	}
> 	set_pseudofs_security(&eep);
> 	exp =3D export_create(&eep, 0);
> 	if (exp =3D=3D NULL)
> --=20
> 2.31.1
>=20

--
Chuck Lever