6.4-stable review patch. If anyone has any objections, please let me know.
------------------
From: Chuck Lever chuck.lever@oracle.com
[ Upstream commit 6372e2ee629894433fe6107d7048536a3280a284 ]
The XDR specification in RFC 8881 looks like this:
struct device_addr4 { layouttype4 da_layout_type; opaque da_addr_body<>; };
struct GETDEVICEINFO4resok { device_addr4 gdir_device_addr; bitmap4 gdir_notification; };
union GETDEVICEINFO4res switch (nfsstat4 gdir_status) { case NFS4_OK: GETDEVICEINFO4resok gdir_resok4; case NFS4ERR_TOOSMALL: count4 gdir_mincount; default: void; };
Looking at nfsd4_encode_getdeviceinfo() ....
When the client provides a zero gd_maxcount, then the Linux NFS server implementation encodes the da_layout_type field and then skips the da_addr_body field completely, proceeding directly to encode gdir_notification field.
There does not appear to be an option in the specification to skip encoding da_addr_body. Moreover, Section 18.40.3 says:
If the client wants to just update or turn off notifications, it MAY send a GETDEVICEINFO operation with gdia_maxcount set to zero. In that event, if the device ID is valid, the reply's da_addr_body field of the gdir_device_addr field will be of zero length.
Since the layout drivers are responsible for encoding the da_addr_body field, put this fix inside the ->encode_getdeviceinfo methods.
Fixes: 9cf514ccfacb ("nfsd: implement pNFS operations") Reviewed-by: Christoph Hellwig hch@lst.de Cc: Tom Haynes loghyr@gmail.com Signed-off-by: Chuck Lever chuck.lever@oracle.com Signed-off-by: Sasha Levin sashal@kernel.org --- fs/nfsd/blocklayoutxdr.c | 9 +++++++++ fs/nfsd/flexfilelayoutxdr.c | 9 +++++++++ fs/nfsd/nfs4xdr.c | 25 +++++++++++-------------- 3 files changed, 29 insertions(+), 14 deletions(-)
diff --git a/fs/nfsd/blocklayoutxdr.c b/fs/nfsd/blocklayoutxdr.c index 8e9c1a0f8d380..1ed2f691ebb90 100644 --- a/fs/nfsd/blocklayoutxdr.c +++ b/fs/nfsd/blocklayoutxdr.c @@ -83,6 +83,15 @@ nfsd4_block_encode_getdeviceinfo(struct xdr_stream *xdr, int len = sizeof(__be32), ret, i; __be32 *p;
+ /* + * See paragraph 5 of RFC 8881 S18.40.3. + */ + if (!gdp->gd_maxcount) { + if (xdr_stream_encode_u32(xdr, 0) != XDR_UNIT) + return nfserr_resource; + return nfs_ok; + } + p = xdr_reserve_space(xdr, len + sizeof(__be32)); if (!p) return nfserr_resource; diff --git a/fs/nfsd/flexfilelayoutxdr.c b/fs/nfsd/flexfilelayoutxdr.c index e81d2a5cf381e..bb205328e043d 100644 --- a/fs/nfsd/flexfilelayoutxdr.c +++ b/fs/nfsd/flexfilelayoutxdr.c @@ -85,6 +85,15 @@ nfsd4_ff_encode_getdeviceinfo(struct xdr_stream *xdr, int addr_len; __be32 *p;
+ /* + * See paragraph 5 of RFC 8881 S18.40.3. + */ + if (!gdp->gd_maxcount) { + if (xdr_stream_encode_u32(xdr, 0) != XDR_UNIT) + return nfserr_resource; + return nfs_ok; + } + /* len + padding for two strings */ addr_len = 16 + da->netaddr.netid_len + da->netaddr.addr_len; ver_len = 20; diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c index ee1a24debd60c..679d9be4abd4d 100644 --- a/fs/nfsd/nfs4xdr.c +++ b/fs/nfsd/nfs4xdr.c @@ -4681,20 +4681,17 @@ nfsd4_encode_getdeviceinfo(struct nfsd4_compoundres *resp, __be32 nfserr,
*p++ = cpu_to_be32(gdev->gd_layout_type);
- /* If maxcount is 0 then just update notifications */ - if (gdev->gd_maxcount != 0) { - ops = nfsd4_layout_ops[gdev->gd_layout_type]; - nfserr = ops->encode_getdeviceinfo(xdr, gdev); - if (nfserr) { - /* - * We don't bother to burden the layout drivers with - * enforcing gd_maxcount, just tell the client to - * come back with a bigger buffer if it's not enough. - */ - if (xdr->buf->len + 4 > gdev->gd_maxcount) - goto toosmall; - return nfserr; - } + ops = nfsd4_layout_ops[gdev->gd_layout_type]; + nfserr = ops->encode_getdeviceinfo(xdr, gdev); + if (nfserr) { + /* + * We don't bother to burden the layout drivers with + * enforcing gd_maxcount, just tell the client to + * come back with a bigger buffer if it's not enough. + */ + if (xdr->buf->len + 4 > gdev->gd_maxcount) + goto toosmall; + return nfserr; }
if (gdev->gd_notify_types) {