From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754992AbaBKTlO (ORCPT ); Tue, 11 Feb 2014 14:41:14 -0500 Received: from mail.linuxfoundation.org ([140.211.169.12]:40335 "EHLO mail.linuxfoundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754022AbaBKTHk (ORCPT ); Tue, 11 Feb 2014 14:07:40 -0500 From: Greg Kroah-Hartman To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman , stable@vger.kernel.org, Boaz Harrosh Subject: [PATCH 3.10 24/79] ore: Fix wrong math in allocation of per device BIO Date: Tue, 11 Feb 2014 11:05:28 -0800 Message-Id: <20140211184721.627545836@linuxfoundation.org> X-Mailer: git-send-email 1.8.5.1.163.gd7aced9 In-Reply-To: <20140211184720.928667275@linuxfoundation.org> References: <20140211184720.928667275@linuxfoundation.org> User-Agent: quilt/0.61-1 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org 3.10-stable review patch. If anyone has any objections, please let me know. ------------------ From: Boaz Harrosh commit aad560b7f63b495f48a7232fd086c5913a676e6f upstream. At IO preparation we calculate the max pages at each device and allocate a BIO per device of that size. The calculation was wrong on some unaligned corner cases offset/length combination and would make prepare return with -ENOMEM. This would be bad for pnfs-objects that would in that case IO through MDS. And fatal for exofs were it would fail writes with EIO. Fix it by doing the proper math, that will work in all cases. (I ran a test with all possible offset/length combinations this time round). Also when reading we do not need to allocate for the parity units since we jump over them. Also lower the max_io_length to take into account the parity pages so not to allocate BIOs bigger than PAGE_SIZE Signed-off-by: Boaz Harrosh Signed-off-by: Greg Kroah-Hartman --- fs/exofs/ore.c | 37 +++++++++++++++++++++++++------------ include/scsi/osd_ore.h | 1 + 2 files changed, 26 insertions(+), 12 deletions(-) --- a/fs/exofs/ore.c +++ b/fs/exofs/ore.c @@ -103,7 +103,7 @@ int ore_verify_layout(unsigned total_com layout->max_io_length = (BIO_MAX_PAGES_KMALLOC * PAGE_SIZE - layout->stripe_unit) * - layout->group_width; + (layout->group_width - layout->parity); if (layout->parity) { unsigned stripe_length = (layout->group_width - layout->parity) * @@ -286,7 +286,8 @@ int ore_get_rw_state(struct ore_layout if (length) { ore_calc_stripe_info(layout, offset, length, &ios->si); ios->length = ios->si.length; - ios->nr_pages = (ios->length + PAGE_SIZE - 1) / PAGE_SIZE; + ios->nr_pages = ((ios->offset & (PAGE_SIZE - 1)) + + ios->length + PAGE_SIZE - 1) / PAGE_SIZE; if (layout->parity) _ore_post_alloc_raid_stuff(ios); } @@ -536,6 +537,7 @@ void ore_calc_stripe_info(struct ore_lay u64 H = LmodS - G * T; u32 N = div_u64(H, U); + u32 Nlast; /* "H - (N * U)" is just "H % U" so it's bound to u32 */ u32 C = (u32)(H - (N * U)) / stripe_unit + G * group_width; @@ -568,6 +570,10 @@ void ore_calc_stripe_info(struct ore_lay si->length = T - H; if (si->length > length) si->length = length; + + Nlast = div_u64(H + si->length + U - 1, U); + si->maxdevUnits = Nlast - N; + si->M = M; } EXPORT_SYMBOL(ore_calc_stripe_info); @@ -583,13 +589,16 @@ int _ore_add_stripe_unit(struct ore_io_s int ret; if (per_dev->bio == NULL) { - unsigned pages_in_stripe = ios->layout->group_width * - (ios->layout->stripe_unit / PAGE_SIZE); - unsigned nr_pages = ios->nr_pages * ios->layout->group_width / - (ios->layout->group_width - - ios->layout->parity); - unsigned bio_size = (nr_pages + pages_in_stripe) / - ios->layout->group_width; + unsigned bio_size; + + if (!ios->reading) { + bio_size = ios->si.maxdevUnits; + } else { + bio_size = (ios->si.maxdevUnits + 1) * + (ios->layout->group_width - ios->layout->parity) / + ios->layout->group_width; + } + bio_size *= (ios->layout->stripe_unit / PAGE_SIZE); per_dev->bio = bio_kmalloc(GFP_KERNEL, bio_size); if (unlikely(!per_dev->bio)) { @@ -609,8 +618,12 @@ int _ore_add_stripe_unit(struct ore_io_s added_len = bio_add_pc_page(q, per_dev->bio, pages[pg], pglen, pgbase); if (unlikely(pglen != added_len)) { - ORE_DBGMSG("Failed bio_add_pc_page bi_vcnt=%u\n", - per_dev->bio->bi_vcnt); + /* If bi_vcnt == bi_max then this is a SW BUG */ + ORE_DBGMSG("Failed bio_add_pc_page bi_vcnt=0x%x " + "bi_max=0x%x BIO_MAX=0x%x cur_len=0x%x\n", + per_dev->bio->bi_vcnt, + per_dev->bio->bi_max_vecs, + BIO_MAX_PAGES_KMALLOC, cur_len); ret = -ENOMEM; goto out; } @@ -1098,7 +1111,7 @@ int ore_truncate(struct ore_layout *layo size_attr->attr = g_attr_logical_length; size_attr->attr.val_ptr = &size_attr->newsize; - ORE_DBGMSG("trunc(0x%llx) obj_offset=0x%llx dev=%d\n", + ORE_DBGMSG2("trunc(0x%llx) obj_offset=0x%llx dev=%d\n", _LLU(oc->comps->obj.id), _LLU(obj_size), i); ret = _truncate_mirrors(ios, i * ios->layout->mirrors_p1, &size_attr->attr); --- a/include/scsi/osd_ore.h +++ b/include/scsi/osd_ore.h @@ -102,6 +102,7 @@ struct ore_striping_info { unsigned unit_off; unsigned cur_pg; unsigned cur_comp; + unsigned maxdevUnits; }; struct ore_io_state;