Re: [PATCH v3 3/6] NFSD: Add READ_PLUS data support

[Chuck Lever <chuck.lever@oracle.com>]

Hi Anna-

> On Aug 3, 2020, at 12:59 PM, schumaker.anna@gmail.com wrote:
> 
> From: Anna Schumaker <Anna.Schumaker@Netapp.com>
> 
> This patch adds READ_PLUS support for returning a single
> NFS4_CONTENT_DATA segment to the client. This is basically the same as
> the READ operation, only with the extra information about data segments.
> 
> Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
> ---
> fs/nfsd/nfs4proc.c | 17 ++++++++++
> fs/nfsd/nfs4xdr.c  | 85 ++++++++++++++++++++++++++++++++++++++++++++--
> 2 files changed, 100 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
> index a09c35f0f6f0..9630d33211f2 100644
> --- a/fs/nfsd/nfs4proc.c
> +++ b/fs/nfsd/nfs4proc.c
> @@ -2523,6 +2523,16 @@ static inline u32 nfsd4_read_rsize(struct svc_rqst *rqstp, struct nfsd4_op *op)
> 	return (op_encode_hdr_size + 2 + XDR_QUADLEN(rlen)) * sizeof(__be32);
> }
> 
> +static inline u32 nfsd4_read_plus_rsize(struct svc_rqst *rqstp, struct nfsd4_op *op)
> +{
> +	u32 maxcount = svc_max_payload(rqstp);
> +	u32 rlen = min(op->u.read.rd_length, maxcount);
> +	/* enough extra xdr space for encoding either a hole or data segment. */
> +	u32 segments = 1 + 2 + 2;
> +
> +	return (op_encode_hdr_size + 2 + segments + XDR_QUADLEN(rlen)) * sizeof(__be32);
> +}
> +
> static inline u32 nfsd4_readdir_rsize(struct svc_rqst *rqstp, struct nfsd4_op *op)
> {
> 	u32 maxcount = 0, rlen = 0;
> @@ -3059,6 +3069,13 @@ static const struct nfsd4_operation nfsd4_ops[] = {
> 		.op_name = "OP_COPY",
> 		.op_rsize_bop = nfsd4_copy_rsize,
> 	},
> +	[OP_READ_PLUS] = {
> +		.op_func = nfsd4_read,
> +		.op_release = nfsd4_read_release,
> +		.op_name = "OP_READ_PLUS",
> +		.op_rsize_bop = nfsd4_read_plus_rsize,
> +		.op_get_currentstateid = nfsd4_get_readstateid,
> +	},
> 	[OP_SEEK] = {
> 		.op_func = nfsd4_seek,
> 		.op_name = "OP_SEEK",
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index 6a1c0a7fae05..1d143bf02b83 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -1957,7 +1957,7 @@ static const nfsd4_dec nfsd4_dec_ops[] = {
> 	[OP_LAYOUTSTATS]	= (nfsd4_dec)nfsd4_decode_notsupp,
> 	[OP_OFFLOAD_CANCEL]	= (nfsd4_dec)nfsd4_decode_offload_status,
> 	[OP_OFFLOAD_STATUS]	= (nfsd4_dec)nfsd4_decode_offload_status,
> -	[OP_READ_PLUS]		= (nfsd4_dec)nfsd4_decode_notsupp,
> +	[OP_READ_PLUS]		= (nfsd4_dec)nfsd4_decode_read,
> 	[OP_SEEK]		= (nfsd4_dec)nfsd4_decode_seek,
> 	[OP_WRITE_SAME]		= (nfsd4_dec)nfsd4_decode_notsupp,
> 	[OP_CLONE]		= (nfsd4_dec)nfsd4_decode_clone,
> @@ -4367,6 +4367,87 @@ nfsd4_encode_offload_status(struct nfsd4_compoundres *resp, __be32 nfserr,
> 		return nfserr_resource;
> 	p = xdr_encode_hyper(p, os->count);
> 	*p++ = cpu_to_be32(0);
> +	return nfserr;
> +}
> +
> +static __be32
> +nfsd4_encode_read_plus_data(struct nfsd4_compoundres *resp,
> +			    struct nfsd4_read *read,
> +			    unsigned long maxcount,  u32 *eof)
> +{
> +	struct xdr_stream *xdr = &resp->xdr;
> +	struct file *file = read->rd_nf->nf_file;
> +	int starting_len = xdr->buf->len;
> +	__be32 nfserr;
> +	__be32 *p, tmp;
> +	__be64 tmp64;
> +
> +	/* Content type, offset, byte count */
> +	p = xdr_reserve_space(xdr, 4 + 8 + 4);
> +	if (!p)
> +		return nfserr_resource;
> +
> +	read->rd_vlen = xdr_reserve_space_vec(xdr, resp->rqstp->rq_vec, maxcount);
> +	if (read->rd_vlen < 0)
> +		return nfserr_resource;
> +
> +	nfserr = nfsd_readv(resp->rqstp, read->rd_fhp, file, read->rd_offset,
> +			    resp->rqstp->rq_vec, read->rd_vlen, &maxcount, eof);

nfsd4_encode_read() has this:

3904         if (file->f_op->splice_read &&
3905             test_bit(RQ_SPLICE_OK, &resp->rqstp->rq_flags))
3906                 nfserr = nfsd4_encode_splice_read(resp, read, file, maxcount);
3907         else
3908                 nfserr = nfsd4_encode_readv(resp, read, file, maxcount);

So READ_PLUS never uses ->splice_read. 

readv is known to be less efficient than splice. Is there a measurable
server performance impact (either latency or CPU utilization) when
reading a file with no holes?


> +	if (nfserr)
> +		return nfserr;
> +	if (svc_encode_read_payload(resp->rqstp, starting_len + 16, maxcount))
> +		return nfserr_io;

Not sure you want a read_payload call here. This is to notify the
transport that there is a section of the message that can be sent
via RDMA, but READ_PLUS has no DDP-eligible data items; especially
not holes!

Although, the call is not likely to be much more than a no-op,
since clients won't be providing any Write chunks for READ_PLUS.


> +
> +	tmp = htonl(NFS4_CONTENT_DATA);
> +	write_bytes_to_xdr_buf(xdr->buf, starting_len,      &tmp,   4);
> +	tmp64 = cpu_to_be64(read->rd_offset);
> +	write_bytes_to_xdr_buf(xdr->buf, starting_len + 4,  &tmp64, 8);
> +	tmp = htonl(maxcount);
> +	write_bytes_to_xdr_buf(xdr->buf, starting_len + 12, &tmp,   4);
> +	return nfs_ok;
> +}
> +
> +static __be32
> +nfsd4_encode_read_plus(struct nfsd4_compoundres *resp, __be32 nfserr,
> +		       struct nfsd4_read *read)
> +{
> +	unsigned long maxcount;
> +	struct xdr_stream *xdr = &resp->xdr;
> +	struct file *file;
> +	int starting_len = xdr->buf->len;
> +	int segments = 0;
> +	__be32 *p, tmp;
> +	u32 eof;
> +
> +	if (nfserr)
> +		return nfserr;
> +	file = read->rd_nf->nf_file;
> +
> +	/* eof flag, segment count */
> +	p = xdr_reserve_space(xdr, 4 + 4);
> +	if (!p)
> +		return nfserr_resource;
> +	xdr_commit_encode(xdr);
> +
> +	maxcount = svc_max_payload(resp->rqstp);
> +	maxcount = min_t(unsigned long, maxcount,
> +			 (xdr->buf->buflen - xdr->buf->len));
> +	maxcount = min_t(unsigned long, maxcount, read->rd_length);
> +
> +	eof = read->rd_offset >= i_size_read(file_inode(file));
> +	if (!eof) {
> +		nfserr = nfsd4_encode_read_plus_data(resp, read, maxcount, &eof);
> +		segments++;
> +	}
> +
> +	if (nfserr)
> +		xdr_truncate_encode(xdr, starting_len);
> +	else {
> +		tmp = htonl(eof);
> +		write_bytes_to_xdr_buf(xdr->buf, starting_len,     &tmp, 4);
> +		tmp = htonl(segments);
> +		write_bytes_to_xdr_buf(xdr->buf, starting_len + 4, &tmp, 4);
> +	}
> 
> 	return nfserr;
> }
> @@ -4509,7 +4590,7 @@ static const nfsd4_enc nfsd4_enc_ops[] = {
> 	[OP_LAYOUTSTATS]	= (nfsd4_enc)nfsd4_encode_noop,
> 	[OP_OFFLOAD_CANCEL]	= (nfsd4_enc)nfsd4_encode_noop,
> 	[OP_OFFLOAD_STATUS]	= (nfsd4_enc)nfsd4_encode_offload_status,
> -	[OP_READ_PLUS]		= (nfsd4_enc)nfsd4_encode_noop,
> +	[OP_READ_PLUS]		= (nfsd4_enc)nfsd4_encode_read_plus,
> 	[OP_SEEK]		= (nfsd4_enc)nfsd4_encode_seek,
> 	[OP_WRITE_SAME]		= (nfsd4_enc)nfsd4_encode_noop,
> 	[OP_CLONE]		= (nfsd4_enc)nfsd4_encode_noop,
> -- 
> 2.27.0
> 

--
Chuck Lever