Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S940336AbdDTFr3 (ORCPT ); Thu, 20 Apr 2017 01:47:29 -0400 Received: from mx2.suse.de ([195.135.220.15]:40386 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S965408AbdDTFp0 (ORCPT ); Thu, 20 Apr 2017 01:45:26 -0400 From: NeilBrown To: Jens Axboe Date: Thu, 20 Apr 2017 15:38:49 +1000 Subject: [PATCH 04/11] block: Improvements to bounce-buffer handling Cc: linux-block@vger.kernel.org, linux-kernel@vger.kernel.org Message-ID: <149266672902.27388.15322863417005890204.stgit@noble> In-Reply-To: <149266645258.27388.14083229348123176454.stgit@noble> References: <149266645258.27388.14083229348123176454.stgit@noble> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3598 Lines: 106 Since commit 23688bf4f830 ("block: ensure to split after potentially bouncing a bio") blk_queue_bounce() is called *before* blk_queue_split(). This means that: 1/ the comments blk_queue_split() about bounce buffers are irrelevant, and 2/ a very large bio (more than BIO_MAX_PAGES) will no longer be split before it arrives at blk_queue_bounce(), leading to the possibility that bio_clone_bioset() will fail and a NULL will be dereferenced. Separately, blk_queue_bounce() shouldn't use fs_bio_set as the bio being copied could be from the same set, and this could lead to a deadlock. So: - allocate 2 private biosets for blk_queue_bounce, one for splitting enormous bios and one for cloning bios. - add code to split a bio that exceeds BIO_MAX_PAGES. - Fix up the comments in blk_queue_split() Signed-off-by: NeilBrown --- block/blk-merge.c | 14 ++++---------- block/bounce.c | 27 ++++++++++++++++++++++++++- 2 files changed, 30 insertions(+), 11 deletions(-) diff --git a/block/blk-merge.c b/block/blk-merge.c index d59074556703..51c84540d3bb 100644 --- a/block/blk-merge.c +++ b/block/blk-merge.c @@ -117,17 +117,11 @@ static struct bio *blk_bio_segment_split(struct request_queue *q, * each holds at most BIO_MAX_PAGES bvecs because * bio_clone() can fail to allocate big bvecs. * - * It should have been better to apply the limit per - * request queue in which bio_clone() is involved, - * instead of globally. The biggest blocker is the - * bio_clone() in bio bounce. + * Those drivers which will need to use bio_clone() + * should tell us in some way. For now, impose the + * BIO_MAX_PAGES limit on all queues. * - * If bio is splitted by this reason, we should have - * allowed to continue bios merging, but don't do - * that now for making the change simple. - * - * TODO: deal with bio bounce's bio_clone() gracefully - * and convert the global limit into per-queue limit. + * TODO: handle users of bio_clone() differently. */ if (bvecs++ >= BIO_MAX_PAGES) goto split; diff --git a/block/bounce.c b/block/bounce.c index 1cb5dd3a5da1..51fb538b504d 100644 --- a/block/bounce.c +++ b/block/bounce.c @@ -26,6 +26,7 @@ #define POOL_SIZE 64 #define ISA_POOL_SIZE 16 +struct bio_set *bounce_bio_set, *bounce_bio_split; static mempool_t *page_pool, *isa_page_pool; #if defined(CONFIG_HIGHMEM) || defined(CONFIG_NEED_BOUNCE_POOL) @@ -40,6 +41,14 @@ static __init int init_emergency_pool(void) BUG_ON(!page_pool); pr_info("pool size: %d pages\n", POOL_SIZE); + bounce_bio_set = bioset_create(BIO_POOL_SIZE, 0); + BUG_ON(!bounce_bio_set); + if (bioset_integrity_create(bounce_bio_set, BIO_POOL_SIZE)) + BUG_ON(1); + + bounce_bio_split = bioset_create_nobvec(BIO_POOL_SIZE, 0); + BUG_ON(!bounce_bio_split); + return 0; } @@ -194,7 +203,23 @@ static void __blk_queue_bounce(struct request_queue *q, struct bio **bio_orig, return; bounce: - bio = bio_clone_bioset(*bio_orig, GFP_NOIO, fs_bio_set); + if (bio_segments(*bio_orig) > BIO_MAX_PAGES) { + int cnt = 0; + int sectors = 0; + struct bio_vec bv; + struct bvec_iter iter; + bio_for_each_segment(bv, *bio_orig, iter) { + if (cnt++ < BIO_MAX_PAGES) + sectors += bv.bv_len >> 9; + else + break; + } + bio = bio_split(*bio_orig, sectors, GFP_NOIO, bounce_bio_split); + bio_chain(bio, *bio_orig); + generic_make_request(*bio_orig); + *bio_orig = bio; + } + bio = bio_clone_bioset(*bio_orig, GFP_NOIO, bounce_bio_set); bio_for_each_segment_all(to, bio, i) { struct page *page = to->bv_page;