leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity

svcrdma: Populate tail iovec when receiving

Browse Source 
So that NFS WRITE payloads can eventually be placed directly into a
file's page cache, enable the RPC-over-RDMA transport to present
these payloads in the xdr_buf's page list, while placing trailing
content (such as a GETATTR operation) in the xdr_buf's tail.

After this change, the RPC-over-RDMA's "copy tail" hack, added by
commit a97c331f9a ("svcrdma: Handle additional inline content"),
is no longer needed and can be removed.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
This commit is contained in:
Chuck Lever 2017-08-18 11:12:35 -04:00 committed by J. Bruce Fields
parent eae03e2ac8
commit 193bcb7b37
1 changed files with 15 additions and 93 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

106
net/sunrpc/xprtrdma/svc_rdma_rw.c
View File

@ -691,78 +691,6 @@ static int svc_rdma_build_read_chunk(struct svc_rqst *rqstp,
return ret; return ret;
} }
/* If there is inline content following the Read chunk, append it to
* the page list immediately following the data payload. This has to
* be done after the reader function has determined how many pages
* were consumed for RDMA Read.
*
* On entry, ri_pageno and ri_pageoff point directly to the end of the
* page list. On exit, both have been updated to the new "next byte".
*
* Assumptions:
* - Inline content fits entirely in rq_pages[0]
* - Trailing content is only a handful of bytes
*/
static int svc_rdma_copy_tail(struct svc_rqst *rqstp,
struct svc_rdma_read_info *info)
{
struct svc_rdma_op_ctxt *head = info->ri_readctxt;
unsigned int tail_length, remaining;
u8 *srcp, *destp;
/* Assert that all inline content fits in page 0. This is an
* implementation limit, not a protocol limit.
*/
if (head->arg.head[0].iov_len > PAGE_SIZE) {
pr_warn_once("svcrdma: too much trailing inline content\n");
return -EINVAL;
}
srcp = head->arg.head[0].iov_base;
srcp += info->ri_position;
tail_length = head->arg.head[0].iov_len - info->ri_position;
remaining = tail_length;
/* If there is room on the last page in the page list, try to
* fit the trailing content there.
*/
if (info->ri_pageoff > 0) {
unsigned int len;
len = min_t(unsigned int, remaining,
PAGE_SIZE - info->ri_pageoff);
destp = page_address(rqstp->rq_pages[info->ri_pageno]);
destp += info->ri_pageoff;
memcpy(destp, srcp, len);
srcp += len;
destp += len;
info->ri_pageoff += len;
remaining -= len;
if (info->ri_pageoff == PAGE_SIZE) {
info->ri_pageno++;
info->ri_pageoff = 0;
}
}
/* Otherwise, a fresh page is needed. */
if (remaining) {
head->arg.pages[info->ri_pageno] =
rqstp->rq_pages[info->ri_pageno];
head->count++;
destp = page_address(rqstp->rq_pages[info->ri_pageno]);
memcpy(destp, srcp, remaining);
info->ri_pageoff += remaining;
}
head->arg.page_len += tail_length;
head->arg.len += tail_length;
head->arg.buflen += tail_length;
return 0;
}
/* Construct RDMA Reads to pull over a normal Read chunk. The chunk /* Construct RDMA Reads to pull over a normal Read chunk. The chunk
* data lands in the page list of head->arg.pages. * data lands in the page list of head->arg.pages.
* *
@ -787,34 +715,28 @@ static int svc_rdma_build_normal_read_chunk(struct svc_rqst *rqstp,
if (ret < 0) if (ret < 0)
goto out; goto out;
/* Read chunk may need XDR round-up (see RFC 5666, s. 3.7). /* Split the Receive buffer between the head and tail
* buffers at Read chunk's position. XDR roundup of the
* chunk is not included in either the pagelist or in
* the tail.
*/ */
if (info->ri_chunklen & 3) { head->arg.tail[0].iov_base =
u32 padlen = 4 - (info->ri_chunklen & 3); head->arg.head[0].iov_base + info->ri_position;
head->arg.tail[0].iov_len =
head->arg.head[0].iov_len - info->ri_position;
head->arg.head[0].iov_len = info->ri_position;
info->ri_chunklen += padlen; /* Read chunk may need XDR roundup (see RFC 5666, s. 3.7).
*
/* NB: data payload always starts on XDR alignment, * NFSv2/3 write decoders need the length of the tail to
* thus the pad can never contain a page boundary. * contain the size of the roundup padding.
*/ */
info->ri_pageoff += padlen; head->arg.tail[0].iov_len += 4 - (info->ri_chunklen & 3);
if (info->ri_pageoff == PAGE_SIZE) {
info->ri_pageno++;
info->ri_pageoff = 0;
}
}
head->arg.page_len = info->ri_chunklen; head->arg.page_len = info->ri_chunklen;
head->arg.len += info->ri_chunklen; head->arg.len += info->ri_chunklen;
head->arg.buflen += info->ri_chunklen; head->arg.buflen += info->ri_chunklen;
if (info->ri_position < head->arg.head[0].iov_len) {
ret = svc_rdma_copy_tail(rqstp, info);
if (ret < 0)
goto out;
}
head->arg.head[0].iov_len = info->ri_position;
out: out:
return ret; return ret;
} }