Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934588AbcJ2IPJ (ORCPT ); Sat, 29 Oct 2016 04:15:09 -0400 Received: from mail-pf0-f194.google.com ([209.85.192.194]:36788 "EHLO mail-pf0-f194.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751824AbcJ2IO4 (ORCPT ); Sat, 29 Oct 2016 04:14:56 -0400 From: Ming Lei To: Jens Axboe , linux-kernel@vger.kernel.org Cc: linux-block@vger.kernel.org, linux-fsdevel@vger.kernel.org, Christoph Hellwig , "Kirill A . Shutemov" , Ming Lei , Jens Axboe Subject: [PATCH 38/60] block: bounce: convert multipage bvecs into singlepage Date: Sat, 29 Oct 2016 16:08:37 +0800 Message-Id: <1477728600-12938-39-git-send-email-tom.leiming@gmail.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1477728600-12938-1-git-send-email-tom.leiming@gmail.com> References: <1477728600-12938-1-git-send-email-tom.leiming@gmail.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2248 Lines: 85 This patch trys to split the incoming multipage bvecs bio, so that the splitted bio can be held into one singlepage bvecs bio. Signed-off-by: Ming Lei --- block/bounce.c | 46 ++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 38 insertions(+), 8 deletions(-) diff --git a/block/bounce.c b/block/bounce.c index a42f7b98b7e6..da240d1de809 100644 --- a/block/bounce.c +++ b/block/bounce.c @@ -187,22 +187,38 @@ static void bounce_end_io_read_isa(struct bio *bio) __bounce_end_io_read(bio, isa_page_pool); } +static inline bool need_split(struct request_queue *q, struct bio *bio) +{ + return bio_sectors(bio) > BIO_SP_MAX_SECTORS; +} + +static inline bool need_bounce(struct request_queue *q, struct bio *bio) +{ + struct bvec_iter iter; + struct bio_vec bv; + + bio_for_each_segment_mp(bv, bio, iter) { + unsigned nr = (bv.bv_offset + bv.bv_len - 1) >> + PAGE_SHIFT; + + if (page_to_pfn(bv.bv_page) + nr > queue_bounce_pfn(q)) + return true; + } + return false; +} + static void __blk_queue_bounce(struct request_queue *q, struct bio **bio_orig, mempool_t *pool) { struct bio *bio; int rw = bio_data_dir(*bio_orig); - struct bio_vec *to, from; - struct bvec_iter iter; + struct bio_vec *to; unsigned i; - bio_for_each_segment(from, *bio_orig, iter) - if (page_to_pfn(from.bv_page) > queue_bounce_pfn(q)) - goto bounce; + if (!need_bounce(q, *bio_orig)) + return; - return; -bounce: - bio = bio_clone_bioset(*bio_orig, GFP_NOIO, fs_bio_set); + bio = bio_clone_bioset_sp(*bio_orig, GFP_NOIO, fs_bio_set); bio_for_each_segment_all(to, bio, i) { struct page *page = to->bv_page; @@ -267,9 +283,23 @@ void blk_queue_bounce(struct request_queue *q, struct bio **bio_orig) pool = isa_page_pool; } + if (!need_bounce(q, *bio_orig)) + return; + /* * slow path + * + * REQ_PC bio won't reach splitting because multipage bvecs + * isn't enabled for REQ_PC. */ + if (need_split(q, *bio_orig)) { + struct bio *split = bio_split(*bio_orig, + BIO_SP_MAX_SECTORS, + GFP_NOIO, q->bio_split); + bio_chain(split, *bio_orig); + generic_make_request(*bio_orig); + *bio_orig = split; + } __blk_queue_bounce(q, bio_orig, pool); } -- 2.7.4