From patchwork Tue Jul 29 21:24:36 2014 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chuck Lever X-Patchwork-Id: 4643051 Return-Path: X-Original-To: patchwork-linux-nfs@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.19.201]) by patchwork1.web.kernel.org (Postfix) with ESMTP id C51BB9F32F for ; Tue, 29 Jul 2014 21:24:43 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id DE09920149 for ; Tue, 29 Jul 2014 21:24:42 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id F144720145 for ; Tue, 29 Jul 2014 21:24:41 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754299AbaG2VYk (ORCPT ); Tue, 29 Jul 2014 17:24:40 -0400 Received: from mail-ig0-f180.google.com ([209.85.213.180]:61097 "EHLO mail-ig0-f180.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754204AbaG2VYj (ORCPT ); Tue, 29 Jul 2014 17:24:39 -0400 Received: by mail-ig0-f180.google.com with SMTP id l13so1732773iga.7 for ; Tue, 29 Jul 2014 14:24:38 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=sender:subject:from:to:cc:date:message-id:in-reply-to:references :user-agent:mime-version:content-type:content-transfer-encoding; bh=fsDrQrMqZzo+gZEDdGEXzjUs2nfHoQWX8ugmc6ahiRA=; b=zqm0NExZdoWwy9cfHCxiiDoEl36QYV+6dcdrkiSHGY/WmCB/qorXuY6Mqklw5G1NTo Qbe7PExiKtLWNqeQnuUeucFogWGuITLdTVmsHrSvjie2E5COQZrNGyHcBXDIqXkZHRXo U4037jbIZRrYoDfTICrtSVVM0IATuiE47bMMchabTe/L8AC9VauouyIiZVox8FzjyRx2 AWf+BdllvlVDeOZorzL+qKGKa9O29YrBJaTyQKKqgJG4CCys8bqu6KWbeAKzSVJpmr8y Q6PQHgiJ1e5HroeRX7Dbbi2OckssJq+XqmwdrXhlgQIYlux1787ATgtvzqn/M3LtrmlA 9Fkw== X-Received: by 10.43.69.202 with SMTP id yd10mr8801045icb.36.1406669078495; Tue, 29 Jul 2014 14:24:38 -0700 (PDT) Received: from manet.1015granger.net ([2604:8800:100:81fc:82ee:73ff:fe43:d64f]) by mx.google.com with ESMTPSA id lq10sm1558322igb.22.2014.07.29.14.24.37 for (version=TLSv1.2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Tue, 29 Jul 2014 14:24:38 -0700 (PDT) Subject: [PATCH v5 10/21] xprtrdma: Properly handle exhaustion of the rb_mws list From: Chuck Lever To: Anna.Schumaker@netapp.com Cc: linux-rdma@vger.kernel.org, linux-nfs@vger.kernel.org Date: Tue, 29 Jul 2014 17:24:36 -0400 Message-ID: <20140729212436.2812.36339.stgit@manet.1015granger.net> In-Reply-To: <20140729211534.2812.3128.stgit@manet.1015granger.net> References: <20140729211534.2812.3128.stgit@manet.1015granger.net> User-Agent: StGit/0.17.1-3-g7d0f MIME-Version: 1.0 Sender: linux-nfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-nfs@vger.kernel.org X-Spam-Status: No, score=-7.5 required=5.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,RP_MATCHES_RCVD,T_DKIM_INVALID,UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP If the rb_mws list is exhausted, clean up and return NULL so that call_allocate() will delay and try again. Signed-off-by: Chuck Lever Tested-by: Steve Wise Tested-by: Shirley Ma Tested-by: Devesh Sharma --- net/sunrpc/xprtrdma/verbs.c | 103 ++++++++++++++++++++++++++++++------------- 1 file changed, 71 insertions(+), 32 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index dd43796..addda7f 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -1256,6 +1256,67 @@ rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf) kfree(buf->rb_pool); } +/* "*mw" can be NULL when rpcrdma_buffer_get_mrs() fails, leaving + * some req segments uninitialized. + */ +static void +rpcrdma_buffer_put_mr(struct rpcrdma_mw **mw, struct rpcrdma_buffer *buf) +{ + if (*mw) { + list_add_tail(&(*mw)->mw_list, &buf->rb_mws); + *mw = NULL; + } +} + +/* Cycle mw's back in reverse order, and "spin" them. + * This delays and scrambles reuse as much as possible. + */ +static void +rpcrdma_buffer_put_mrs(struct rpcrdma_req *req, struct rpcrdma_buffer *buf) +{ + struct rpcrdma_mr_seg *seg = req->rl_segments; + struct rpcrdma_mr_seg *seg1 = seg; + int i; + + for (i = 1, seg++; i < RPCRDMA_MAX_SEGS; seg++, i++) + rpcrdma_buffer_put_mr(&seg->mr_chunk.rl_mw, buf); + rpcrdma_buffer_put_mr(&seg1->mr_chunk.rl_mw, buf); +} + +static void +rpcrdma_buffer_put_sendbuf(struct rpcrdma_req *req, struct rpcrdma_buffer *buf) +{ + buf->rb_send_bufs[--buf->rb_send_index] = req; + req->rl_niovs = 0; + if (req->rl_reply) { + buf->rb_recv_bufs[--buf->rb_recv_index] = req->rl_reply; + req->rl_reply->rr_func = NULL; + req->rl_reply = NULL; + } +} + +static struct rpcrdma_req * +rpcrdma_buffer_get_mrs(struct rpcrdma_req *req, struct rpcrdma_buffer *buf) +{ + struct rpcrdma_mw *r; + int i; + + i = RPCRDMA_MAX_SEGS - 1; + while (!list_empty(&buf->rb_mws)) { + r = list_entry(buf->rb_mws.next, + struct rpcrdma_mw, mw_list); + list_del(&r->mw_list); + req->rl_segments[i].mr_chunk.rl_mw = r; + if (unlikely(i-- == 0)) + return req; /* Success */ + } + + /* Not enough entries on rb_mws for this req */ + rpcrdma_buffer_put_sendbuf(req, buf); + rpcrdma_buffer_put_mrs(req, buf); + return NULL; +} + /* * Get a set of request/reply buffers. * @@ -1268,10 +1329,9 @@ rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf) struct rpcrdma_req * rpcrdma_buffer_get(struct rpcrdma_buffer *buffers) { + struct rpcrdma_ia *ia = rdmab_to_ia(buffers); struct rpcrdma_req *req; unsigned long flags; - int i; - struct rpcrdma_mw *r; spin_lock_irqsave(&buffers->rb_lock, flags); if (buffers->rb_send_index == buffers->rb_max_requests) { @@ -1291,14 +1351,13 @@ rpcrdma_buffer_get(struct rpcrdma_buffer *buffers) buffers->rb_recv_bufs[buffers->rb_recv_index++] = NULL; } buffers->rb_send_bufs[buffers->rb_send_index++] = NULL; - if (!list_empty(&buffers->rb_mws)) { - i = RPCRDMA_MAX_SEGS - 1; - do { - r = list_entry(buffers->rb_mws.next, - struct rpcrdma_mw, mw_list); - list_del(&r->mw_list); - req->rl_segments[i].mr_chunk.rl_mw = r; - } while (--i >= 0); + switch (ia->ri_memreg_strategy) { + case RPCRDMA_FRMR: + case RPCRDMA_MTHCAFMR: + req = rpcrdma_buffer_get_mrs(req, buffers); + break; + default: + break; } spin_unlock_irqrestore(&buffers->rb_lock, flags); return req; @@ -1313,34 +1372,14 @@ rpcrdma_buffer_put(struct rpcrdma_req *req) { struct rpcrdma_buffer *buffers = req->rl_buffer; struct rpcrdma_ia *ia = rdmab_to_ia(buffers); - int i; unsigned long flags; spin_lock_irqsave(&buffers->rb_lock, flags); - buffers->rb_send_bufs[--buffers->rb_send_index] = req; - req->rl_niovs = 0; - if (req->rl_reply) { - buffers->rb_recv_bufs[--buffers->rb_recv_index] = req->rl_reply; - req->rl_reply->rr_func = NULL; - req->rl_reply = NULL; - } + rpcrdma_buffer_put_sendbuf(req, buffers); switch (ia->ri_memreg_strategy) { case RPCRDMA_FRMR: case RPCRDMA_MTHCAFMR: - /* - * Cycle mw's back in reverse order, and "spin" them. - * This delays and scrambles reuse as much as possible. - */ - i = 1; - do { - struct rpcrdma_mw **mw; - mw = &req->rl_segments[i].mr_chunk.rl_mw; - list_add_tail(&(*mw)->mw_list, &buffers->rb_mws); - *mw = NULL; - } while (++i < RPCRDMA_MAX_SEGS); - list_add_tail(&req->rl_segments[0].mr_chunk.rl_mw->mw_list, - &buffers->rb_mws); - req->rl_segments[0].mr_chunk.rl_mw = NULL; + rpcrdma_buffer_put_mrs(req, buffers); break; default: break;