From patchwork Thu Apr 23 19:06:31 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Douglas Fuller X-Patchwork-Id: 6264681 Return-Path: X-Original-To: patchwork-ceph-devel@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id 9A5C89F313 for ; Thu, 23 Apr 2015 19:07:10 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 9FDF720380 for ; Thu, 23 Apr 2015 19:07:09 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 92CF42024D for ; Thu, 23 Apr 2015 19:07:08 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030600AbbDWTHF (ORCPT ); Thu, 23 Apr 2015 15:07:05 -0400 Received: from mx1.redhat.com ([209.132.183.28]:53765 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1030449AbbDWTHD (ORCPT ); Thu, 23 Apr 2015 15:07:03 -0400 Received: from int-mx13.intmail.prod.int.phx2.redhat.com (int-mx13.intmail.prod.int.phx2.redhat.com [10.5.11.26]) by mx1.redhat.com (Postfix) with ESMTPS id 95CCE8EA30 for ; Thu, 23 Apr 2015 19:07:03 +0000 (UTC) Received: from rex001.front.sepia.ceph.com (vpn-56-33.rdu2.redhat.com [10.10.56.33]) by int-mx13.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id t3NJ6vZW006593 for ; Thu, 23 Apr 2015 15:07:03 -0400 From: Douglas Fuller To: ceph-devel@vger.kernel.org Subject: [PATCH 1/3] ceph: support multiple class method calls in one ceph_msg Date: Thu, 23 Apr 2015 12:06:31 -0700 Message-Id: In-Reply-To: References: In-Reply-To: References: X-Scanned-By: MIMEDefang 2.68 on 10.5.11.26 Sender: ceph-devel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: ceph-devel@vger.kernel.org X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, T_RP_MATCHES_RCVD, 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 Messages are received from the wire directly into destination buffers. A short read could result in corruption of the following destination buffers. Allocate a single message buffer for all class method calls and split them at the osd_client level. This only applies to ceph_msgs containing multiple op call and may break support for ceph_msgs containing a mix of class method calls that return data and other ops. Signed-off-by: Douglas Fuller --- include/linux/ceph/osd_client.h | 1 + net/ceph/messenger.c | 4 ++ net/ceph/osd_client.c | 92 ++++++++++++++++++++++++++++++++++++++++- 3 files changed, 95 insertions(+), 2 deletions(-) diff --git a/include/linux/ceph/osd_client.h b/include/linux/ceph/osd_client.h index 61b19c4..65fcf80 100644 --- a/include/linux/ceph/osd_client.h +++ b/include/linux/ceph/osd_client.h @@ -99,6 +99,7 @@ struct ceph_osd_req_op { struct ceph_osd_data request_info; struct ceph_osd_data request_data; struct ceph_osd_data response_data; + struct ceph_osd_data chain_data; __u8 class_len; __u8 method_len; __u8 argc; diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c index 967080a..ec04be4 100644 --- a/net/ceph/messenger.c +++ b/net/ceph/messenger.c @@ -907,6 +907,10 @@ static void ceph_msg_data_pages_cursor_init(struct ceph_msg_data_cursor *cursor, BUG_ON(!data->pages); BUG_ON(!data->length); + /* + * bug here if a short read occurs and length < data->length; see + * http://tracker.ceph.com/issues/11424 + */ cursor->resid = min(length, data->length); page_count = calc_pages_for(data->alignment, (u64)data->length); cursor->page_offset = data->alignment & ~PAGE_MASK; diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index 41a4abc..3999dfa 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -301,6 +301,73 @@ static void osd_req_op_data_release(struct ceph_osd_request *osd_req, } } +static int __build_op_cls_chain(struct ceph_osd_request *osd_req) +{ + u64 chain_length = 0; + u32 chain_pagecount = 0; + struct ceph_osd_req_op *op = NULL; + struct ceph_osd_data *osd_data; + struct ceph_osd_data *chain_data; + struct page **pages; + int i; + + chain_data = osd_req_op_data(osd_req, 0, cls, chain_data); + + for (i=0; ir_num_ops; i++) + { + op = &osd_req->r_ops[i]; + if (op->op != CEPH_OSD_OP_CALL) + break; + + osd_data = osd_req_op_data(osd_req, i, cls, chain_data); + osd_data->length = 0; + + osd_data = osd_req_op_data(osd_req, i, cls, response_data); + chain_length += osd_data->length; + } + + chain_data->length = chain_length; + chain_pagecount = (u32)calc_pages_for(0, chain_data->length); + pages = ceph_alloc_page_vector(chain_pagecount, GFP_KERNEL); + if (IS_ERR(pages)) + return PTR_ERR(pages); + ceph_osd_data_pages_init(chain_data, pages, chain_length, 0, false, false); + + return 0; +} + +static int __split_cls_op_chain(struct ceph_osd_request *osd_req) +{ + int i; + void * data; + void * p; + struct ceph_osd_data *osd_data; + + osd_data = osd_req_op_data(osd_req, 0, cls, chain_data); + + if (osd_data->length == 0) + return 0; + + data = kzalloc(osd_data->length, GFP_KERNEL); + if (!data) + return -ENOMEM; + + ceph_copy_from_page_vector(osd_data->pages, data, 0, osd_data->length); + ceph_osd_data_release(osd_data); + + p = data; + for (i = 0; i < osd_req->r_num_ops; i++) + { + osd_data = osd_req_op_data(osd_req, i, cls, response_data); + ceph_copy_to_page_vector(osd_data->pages, p, + 0, osd_req->r_reply_op_len[i]); + p += osd_req->r_reply_op_len[i]; + } + + kfree(data); + return 0; +} + /* * requests */ @@ -694,8 +761,20 @@ static u64 osd_req_encode_op(struct ceph_osd_request *req, src->payload_len += data_length; request_data_len += data_length; } - osd_data = &src->cls.response_data; - ceph_osdc_msg_data_add(req->r_reply, osd_data); + if (which == 0) + { + int err; + + err = __build_op_cls_chain(req); + if (err == -ENOMEM) + { + pr_err("error allocating memory for op chain\n"); + return 0; + } + osd_data = &src->cls.chain_data; + if (osd_data->length) + ceph_osdc_msg_data_add(req->r_reply, osd_data); + } break; case CEPH_OSD_OP_STARTSYNC: break; @@ -1825,6 +1904,15 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg, for (i = 0; i < numops; i++) req->r_reply_op_result[i] = ceph_decode_32(&p); + if (req->r_ops[0].op == CEPH_OSD_OP_CALL && + req->r_ops[0].cls.chain_data.length) + { + int err; + err = __split_cls_op_chain(req); + if (err == -ENOMEM) + goto bad_put; + } + if (le16_to_cpu(msg->hdr.version) >= 6) { p += 8 + 4; /* skip replay_version */ p += 8; /* skip user_version */