linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "J. Bruce Fields" <bfields@redhat.com>
To: linux-nfs@vger.kernel.org
Cc: "J. Bruce Fields" <bfields@redhat.com>
Subject: [PATCH 43/50] nfsd4: turn off zero-copy-read in exotic cases
Date: Sat, 22 Mar 2014 21:12:14 -0400	[thread overview]
Message-ID: <1395537141-10389-44-git-send-email-bfields@redhat.com> (raw)
In-Reply-To: <1395537141-10389-1-git-send-email-bfields@redhat.com>

From: "J. Bruce Fields" <bfields@redhat.com>

We currently allow only one read per compound, with operations before
and after whose responses will require no more than about a page to
encode.

While we don't expect clients to violate those limits any time soon,
this limitation isn't really condoned by the spec, so to future proof
the server we should lift the limitation.

At the same time we'd like to continue to support zero-copy reads.

Supporting multiple zero-copy-reads per compound would require a new
data structure to replace struct xdr_buf, which can represent only one
set of included pages.

So for now we plan to modify encode_read() to support either zero-copy
or non-zero-copy reads, and use some heuristics at the start of the
compound processing to decide whether a zero-copy read will work.

This will allow us to support more exotic compounds without introducing
a performance regression in the normal case.

Later patches handle those "exotic compounds", this one just makes sure
zero-copy is turned off in those cases.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
---
 fs/nfsd/nfs4xdr.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index c96a05e..8cdc346 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -1630,6 +1630,8 @@ nfsd4_decode_compound(struct nfsd4_compoundargs *argp)
 	struct nfsd4_op *op;
 	bool cachethis = false;
 	int max_reply = 2 * RPC_MAX_AUTH_SIZE; /* uh, kind of a guess */
+	int readcount = 0;
+	int readbytes = 0;
 	int i;
 
 	READ_BUF(4);
@@ -1680,14 +1682,21 @@ nfsd4_decode_compound(struct nfsd4_compoundargs *argp)
 		 */
 		cachethis |= nfsd4_cache_this_op(op);
 
-		max_reply += nfsd4_max_reply(argp->rqstp, op);
+		if (op->opnum == OP_READ) {
+			readcount++;
+			readbytes += nfsd4_max_reply(argp->rqstp, op);
+		} else
+			max_reply += nfsd4_max_reply(argp->rqstp, op);
 	}
 	/* Sessions make the DRC unnecessary: */
 	if (argp->minorversion)
 		cachethis = false;
-	svc_reserve(argp->rqstp, max_reply);
+	svc_reserve(argp->rqstp, max_reply + readbytes);
 	argp->rqstp->rq_cachetype = cachethis ? RC_REPLBUFF : RC_NOCACHE;
 
+	if (readcount > 1 || max_reply > PAGE_SIZE - 2*RPC_MAX_AUTH_SIZE)
+		argp->rqstp->rq_splice_ok = false;
+
 	DECODE_TAIL;
 }
 
-- 
1.8.5.3


  parent reply	other threads:[~2014-03-23  1:12 UTC|newest]

Thread overview: 63+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-03-23  1:11 nfsd4 xdr encoding fixes J. Bruce Fields
2014-03-23  1:11 ` [PATCH 01/50] rpc: Allow xdr_buf_subsegment to operate in-place J. Bruce Fields
2014-03-23  1:11 ` [PATCH 02/50] nfsd4: update comments with obsolete function name J. Bruce Fields
2014-03-23  1:11 ` [PATCH 03/50] nfsd4: nfsd4_replay_cache_entry should be static J. Bruce Fields
2014-03-23  1:11 ` [PATCH 04/50] nfsd4: minor nfsd4_replay_cache_entry cleanup J. Bruce Fields
2014-03-23  1:11 ` [PATCH 05/50] nfsd4: use more generous NFS4_ACL_MAX J. Bruce Fields
2014-03-23  1:11 ` [PATCH 06/50] nfsd4: remove redundant check from nfsd4_check_resp_size J. Bruce Fields
2014-03-23  1:11 ` [PATCH 07/50] nfsd4: fix setclientid encode size J. Bruce Fields
2014-03-23  1:11 ` [PATCH 08/50] nfsd4: fix nfs4err_resource in 4.1 case J. Bruce Fields
2014-03-29 19:18   ` J. Bruce Fields
2014-03-23  1:11 ` [PATCH 09/50] nfsd4: embed xdr_stream in nfsd4_compoundres J. Bruce Fields
2014-03-23  1:11 ` [PATCH 10/50] nfsd4: tweak nfsd4_encode_getattr to take xdr_stream J. Bruce Fields
2014-03-23  1:11 ` [PATCH 11/50] nfsd4: move proc_compound xdr encode init to helper J. Bruce Fields
2014-03-23  1:11 ` [PATCH 12/50] nfsd4: reserve head space for krb5 integ/priv info J. Bruce Fields
2014-03-23  1:11 ` [PATCH 13/50] nfsd4: move nfsd4_operation to xdr4.h J. Bruce Fields
2014-03-23  1:11 ` [PATCH 14/50] nfsd4: fix encoding of out-of-space replies J. Bruce Fields
2014-03-23  1:11 ` [PATCH 15/50] nfsd4: allow space for final error return J. Bruce Fields
2014-03-23  1:11 ` [PATCH 16/50] nfsd4: READ, READDIR, etc., are idempotent J. Bruce Fields
2014-03-23  1:11 ` [PATCH 17/50] nfsd4: use xdr_reserve_space in attribute encoding J. Bruce Fields
2014-03-23  1:11 ` [PATCH 18/50] nfsd4: use xdr_stream throughout compound encoding J. Bruce Fields
2014-03-23  6:43   ` Christoph Hellwig
2014-03-23 15:11     ` J. Bruce Fields
2014-03-25 15:38       ` Christoph Hellwig
2014-03-23  1:11 ` [PATCH 19/50] nfsd4: no need for encode_compoundres to adjust lengths J. Bruce Fields
2014-03-23  1:11 ` [PATCH 20/50] nfsd4: keep xdr buf length updated J. Bruce Fields
2014-03-23  6:47   ` Christoph Hellwig
2014-03-23  1:11 ` [PATCH 21/50] rpc: xdr_truncate_encode J. Bruce Fields
2014-03-23  1:11 ` [PATCH 22/50] nfsd4: use xdr_truncate_encode J. Bruce Fields
2014-03-23  6:50   ` Christoph Hellwig
2014-03-23 15:07     ` J. Bruce Fields
2014-03-25 15:36       ` Christoph Hellwig
2014-04-05  0:20         ` J. Bruce Fields
2014-03-23  1:11 ` [PATCH 23/50] nfsd4: "backfill" using write_bytes_to_xdr_buf J. Bruce Fields
2014-03-23  6:51   ` Christoph Hellwig
2014-03-23 14:43     ` J. Bruce Fields
2014-03-23 14:52       ` Christoph Hellwig
2014-03-23  1:11 ` [PATCH 24/50] nfsd4: remove ADJUST_ARGS J. Bruce Fields
2014-03-23  1:11 ` [PATCH 25/50] nfsd4: teach encoders to handle reserve_space failures J. Bruce Fields
2014-03-23  1:11 ` [PATCH 26/50] nfsd4: reserve space before inlining 0-copy pages J. Bruce Fields
2014-03-23  1:11 ` [PATCH 27/50] nfsd4: nfsd4_check_resp_size needn't recalculate length J. Bruce Fields
2014-03-23  1:11 ` [PATCH 28/50] nfsd4: remove redundant encode buffer size checking J. Bruce Fields
2014-03-23  1:12 ` [PATCH 29/50] nfsd4: size-checking cleanup J. Bruce Fields
2014-03-23  1:12 ` [PATCH 30/50] nfsd4: allow encoding across page boundaries J. Bruce Fields
2014-03-23  1:12 ` [PATCH 31/50] nfsd4: convert 4.1 replay encoding J. Bruce Fields
2014-03-23  1:12 ` [PATCH 32/50] nfsd4: don't try to encode conflicting owner if low on space J. Bruce Fields
2014-03-23  1:12 ` [PATCH 33/50] nfsd4: more precise nfsd4_max_reply J. Bruce Fields
2014-03-23  1:12 ` [PATCH 34/50] nfsd4: minor encode_read cleanup J. Bruce Fields
2014-03-23  1:12 ` [PATCH 35/50] nfsd4: nfsd4_check_resp_size should check against whole buffer J. Bruce Fields
2014-03-23  1:12 ` [PATCH 36/50] nfsd4: allow larger 4.1 session drc slots J. Bruce Fields
2014-03-23  1:12 ` [PATCH 37/50] rpc: define xdr_restrict_buflen J. Bruce Fields
2014-03-23  1:12 ` [PATCH 38/50] nfsd4: adjust buflen to session channel limit J. Bruce Fields
2014-03-23  1:12 ` [PATCH 39/50] nfsd4: use session limits to release send buffer reservation J. Bruce Fields
2014-03-23  1:12 ` [PATCH 40/50] nfsd4: allow large readdirs J. Bruce Fields
2014-03-23  1:12 ` [PATCH 41/50] nfsd4: enforce rd_dircount J. Bruce Fields
2014-03-23  1:12 ` [PATCH 42/50] nfsd4: don't treat readlink like a zero-copy operation J. Bruce Fields
2014-03-23  1:12 ` J. Bruce Fields [this message]
2014-03-23  1:12 ` [PATCH 44/50] nfsd4: nfsd_vfs_read doesn't use file handle parameter J. Bruce Fields
2014-03-23  1:12 ` [PATCH 45/50] nfsd4: separate splice and readv cases J. Bruce Fields
2014-03-23  1:12 ` [PATCH 46/50] nfsd4: allow exotic read compounds J. Bruce Fields
2014-03-23  1:12 ` [PATCH 47/50] nfsd4: kill WRITE32 J. Bruce Fields
2014-03-23  1:12 ` [PATCH 48/50] nfsd4: kill WRITE64 J. Bruce Fields
2014-03-23  1:12 ` [PATCH 49/50] nfsd4: kill WRITEMEM J. Bruce Fields
2014-03-23  1:12 ` [PATCH 50/50] nfsd4: kill write32, write64 J. Bruce Fields

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1395537141-10389-44-git-send-email-bfields@redhat.com \
    --to=bfields@redhat.com \
    --cc=linux-nfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).