diff options
| author | Chuck Lever <chuck.lever@oracle.com> | 2016-03-01 13:06:02 -0500 |
|---|---|---|
| committer | J. Bruce Fields <bfields@redhat.com> | 2016-03-01 13:06:35 -0800 |
| commit | 4500632f60fa0d85e4101c374898cdf9b7b0cfac (patch) | |
| tree | 108e40fab78bcb843b2a7efa9e6439f8f2d4249b /fs/nfsd | |
| parent | f6763c29ab86c3ee27760a06e07bbeab47635b61 (diff) | |
| download | linux-4500632f60fa0d85e4101c374898cdf9b7b0cfac.tar.gz linux-4500632f60fa0d85e4101c374898cdf9b7b0cfac.tar.bz2 linux-4500632f60fa0d85e4101c374898cdf9b7b0cfac.zip | |
nfsd: Lower NFSv4.1 callback message size limit
The maximum size of a backchannel message on RPC-over-RDMA depends
on the connection's inline threshold. Today that threshold is
typically 1024 bytes, making the maximum message size 996 bytes.
The Linux server's CREATE_SESSION operation checks that the size
of callback Calls can be as large as 1044 bytes, to accommodate
RPCSEC_GSS. Thus CREATE_SESSION fails if a client advertises the
true message size maximum of 996 bytes.
But the server's backchannel currently does not support RPCSEC_GSS.
The actual maximum size it needs is much smaller. It is safe to
reduce the limit to enable NFSv4.1 on RDMA backchannel operation.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Diffstat (limited to 'fs/nfsd')
| -rw-r--r-- | fs/nfsd/nfs4state.c | 21 |
1 files changed, 13 insertions, 8 deletions
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index d4b0b789ea24..0462eeddfff9 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -2587,21 +2587,26 @@ static __be32 check_forechannel_attrs(struct nfsd4_channel_attrs *ca, struct nfs return nfs_ok; } +/* + * Server's NFSv4.1 backchannel support is AUTH_SYS-only for now. + * These are based on similar macros in linux/sunrpc/msg_prot.h . + */ +#define RPC_MAX_HEADER_WITH_AUTH_SYS \ + (RPC_CALLHDRSIZE + 2 * (2 + UNX_CALLSLACK)) + +#define RPC_MAX_REPHEADER_WITH_AUTH_SYS \ + (RPC_REPHDRSIZE + (2 + NUL_REPLYSLACK)) + #define NFSD_CB_MAX_REQ_SZ ((NFS4_enc_cb_recall_sz + \ - RPC_MAX_HEADER_WITH_AUTH) * sizeof(__be32)) + RPC_MAX_HEADER_WITH_AUTH_SYS) * sizeof(__be32)) #define NFSD_CB_MAX_RESP_SZ ((NFS4_dec_cb_recall_sz + \ - RPC_MAX_REPHEADER_WITH_AUTH) * sizeof(__be32)) + RPC_MAX_REPHEADER_WITH_AUTH_SYS) * \ + sizeof(__be32)) static __be32 check_backchannel_attrs(struct nfsd4_channel_attrs *ca) { ca->headerpadsz = 0; - /* - * These RPC_MAX_HEADER macros are overkill, especially since we - * don't even do gss on the backchannel yet. But this is still - * less than 1k. Tighten up this estimate in the unlikely event - * it turns out to be a problem for some client: - */ if (ca->maxreq_sz < NFSD_CB_MAX_REQ_SZ) return nfserr_toosmall; if (ca->maxresp_sz < NFSD_CB_MAX_RESP_SZ) |