linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Chuck Lever <chuck.lever@oracle.com>
To: linux-rdma@vger.kernel.org, linux-nfs@vger.kernel.org
Subject: [PATCH v1 14/19] xprtrdma: More Send completion batching
Date: Wed, 10 Apr 2019 16:07:46 -0400	[thread overview]
Message-ID: <20190410200746.11522.64034.stgit@manet.1015granger.net> (raw)
In-Reply-To: <20190410200446.11522.21145.stgit@manet.1015granger.net>

Instead of using a fixed number, allow the amount of Send completion
batching to vary based on the client's maximum credit limit.

- A larger default gives a small boost to IOPS throughput

- Reducing it based on max_requests gives a safe result when the
  max credit limit is cranked down (eg. when the device has a small
  max_qp_wr).

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 net/sunrpc/xprtrdma/verbs.c     |    4 +---
 net/sunrpc/xprtrdma/xprt_rdma.h |   10 ----------
 2 files changed, 1 insertion(+), 13 deletions(-)

diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index 378b460..f65d17a 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -521,9 +521,7 @@ static void rpcrdma_xprt_drain(struct rpcrdma_xprt *r_xprt)
 		ep->rep_attr.cap.max_send_sge,
 		ep->rep_attr.cap.max_recv_sge);
 
-	/* set trigger for requesting send completion */
-	ep->rep_send_batch = min_t(unsigned int, RPCRDMA_MAX_SEND_BATCH,
-				   cdata->max_requests >> 2);
+	ep->rep_send_batch = cdata->max_requests >> 3;
 	ep->rep_send_count = ep->rep_send_batch;
 	init_waitqueue_head(&ep->rep_connect_wait);
 	ep->rep_receive_count = 0;
diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h
index 7d82c91..5063fa7 100644
--- a/net/sunrpc/xprtrdma/xprt_rdma.h
+++ b/net/sunrpc/xprtrdma/xprt_rdma.h
@@ -232,16 +232,6 @@ struct rpcrdma_sendctx {
 	struct ib_sge		sc_sges[];
 };
 
-/* Limit the number of SGEs that can be unmapped during one
- * Send completion. This caps the amount of work a single
- * completion can do before returning to the provider.
- *
- * Setting this to zero disables Send completion batching.
- */
-enum {
-	RPCRDMA_MAX_SEND_BATCH = 7,
-};
-
 /*
  * struct rpcrdma_mr - external memory region metadata
  *


  parent reply	other threads:[~2019-04-10 20:07 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-10 20:06 [PATCH v1 00/19] Proposed NFS/RDMA patches for v5.2 Chuck Lever
2019-04-10 20:06 ` [PATCH v1 01/19] SUNRPC: Avoid digging into the ATOMIC pool Chuck Lever
2019-04-10 20:06 ` [PATCH v1 02/19] xprtrdma: Fix an frwr_map recovery nit Chuck Lever
2019-04-10 20:06 ` [PATCH v1 03/19] xprtrdma: Defer completion only when local invalidation is needed Chuck Lever
2019-04-10 20:06 ` [PATCH v1 04/19] xprtrdma: Clean up rpcrdma_create_req() Chuck Lever
2019-04-10 20:06 ` [PATCH v1 05/19] xprtrdma: Clean up rpcrdma_create_rep() and rpcrdma_destroy_rep() Chuck Lever
2019-04-11 20:47   ` Anna Schumaker
2019-04-10 20:07 ` [PATCH v1 06/19] xprtrdma: rpcrdma_regbuf alignment Chuck Lever
2019-04-10 20:07 ` [PATCH v1 07/19] xprtrdma: Allocate req's regbufs at xprt create time Chuck Lever
2019-04-10 20:07 ` [PATCH v1 08/19] xprtrdma: De-duplicate "allocate new, free old regbuf" Chuck Lever
2019-04-10 20:07 ` [PATCH v1 09/19] xprtrdma: Clean up regbuf helpers Chuck Lever
2019-04-10 20:07 ` [PATCH v1 10/19] xprtrdma: Backchannel can use GFP_KERNEL allocations Chuck Lever
2019-04-10 20:07 ` [PATCH v1 11/19] xprtrdma: Increase maximum number of backchannel request Chuck Lever
2019-04-10 20:07 ` [PATCH v1 12/19] xprtrdma: Trace marshaling failures Chuck Lever
2019-04-10 20:07 ` [PATCH v1 13/19] xprtrdma: Clean up sendctx functions Chuck Lever
2019-04-10 20:07 ` Chuck Lever [this message]
2019-04-10 20:07 ` [PATCH v1 15/19] xprtrdma: Eliminate rpcrdma_ia::ri_device Chuck Lever
2019-04-10 20:07 ` [PATCH v1 16/19] SUNRPC: Update comments based on recent changes Chuck Lever
2019-04-10 20:08 ` [PATCH v1 17/19] xprtrdma: Remove rpcrdma_create_data_internal::rsize and wsize Chuck Lever
2019-04-10 20:08 ` [PATCH v1 18/19] xprtrdma: Aggregate the inline settings in struct rpcrdma_ep Chuck Lever
2019-04-10 20:08 ` [PATCH v1 19/19] xprtrdma: Eliminate struct rpcrdma_create_data_internal Chuck Lever

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=20190410200746.11522.64034.stgit@manet.1015granger.net \
    --to=chuck.lever@oracle.com \
    --cc=linux-nfs@vger.kernel.org \
    --cc=linux-rdma@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).