From patchwork Tue Oct 27 14:55:16 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: =?utf-8?q?Matias_Bj=C3=B8rling?= X-Patchwork-Id: 7497421 Return-Path: X-Original-To: patchwork-linux-fsdevel@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork2.web.kernel.org (Postfix) with ESMTP id A2531BF90C for ; Tue, 27 Oct 2015 14:58:37 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 3380720918 for ; Tue, 27 Oct 2015 14:58:36 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 7A84F2091D for ; Tue, 27 Oct 2015 14:58:34 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932795AbbJ0O5o (ORCPT ); Tue, 27 Oct 2015 10:57:44 -0400 Received: from mail-wi0-f169.google.com ([209.85.212.169]:36761 "EHLO mail-wi0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932677AbbJ0Ozg (ORCPT ); Tue, 27 Oct 2015 10:55:36 -0400 Received: by wicfx6 with SMTP id fx6so163378360wic.1 for ; Tue, 27 Oct 2015 07:55:34 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=bjorling.me; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-type:content-transfer-encoding; bh=WfFE4nMyR2+tVoyajnJpudDEo9zBLjNXLI1LQUU/sEk=; b=G8qoUk10CkStKsyhI0nJ7NWjrbQlH4N+4HvhfFYc9P0ggSrjFsGEcBrOTaL3wxsfLZ yjy8XZRzr+TPTeqhPTRmrpQX27rREo0IABwNXW873uJ39nVJIbBess/2T1J+F47TXAtd sj7naGTodB6AQo/4gDwgtzzyRmQo9lCl/Ocrg= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-type:content-transfer-encoding; bh=WfFE4nMyR2+tVoyajnJpudDEo9zBLjNXLI1LQUU/sEk=; b=b0RKAAr9SipiIT1ZQumsUjbFgB4FFmqF5sgrGt7/vAaXfqtpQDm2VYaI+sD2kLC9LH qIl425/MNDTXKpqNIpshvOGQ3ZwWxQKdc5cF4JjJ7J5B733cZ6Md+R9SGj2gy5t/hGd5 7Q9L4mKETLcxgbdoNuaEo2Futd03uIXQhyOsB3/mgiUjz4jRicPha+ky8YPOfMktbUud 0byueSsqH9rxCUEItAQItqkdQCs1pEXWOwmF/h29ynsrzQIV+kkLIXUl0EnWm//8p6lG SL6Ul78Wcvvx/0A4VgHYXbXXOWkk0qtXDyVsHTgIzBvtUbc1sO1HXkpyqJkqUtR7doQf 3ixw== X-Gm-Message-State: ALoCoQmp+doJRjoqL28MQUCGi6Cwfumq+PerBf19Z5Gh/5Wm3d/qrhZqutJ1XyYzIQFhmoAXl94V X-Received: by 10.180.102.230 with SMTP id fr6mr25719809wib.66.1445957734209; Tue, 27 Oct 2015 07:55:34 -0700 (PDT) Received: from localhost.localdomain (6164198-cl69.boa.fiberby.dk. [193.106.164.198]) by smtp.gmail.com with ESMTPSA id jh4sm40689536wjb.33.2015.10.27.07.55.32 (version=TLSv1.2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Tue, 27 Oct 2015 07:55:32 -0700 (PDT) From: =?UTF-8?q?Matias=20Bj=C3=B8rling?= To: axboe@fb.com, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, linux-nvme@lists.infradead.org Cc: hch@infradead.org, jg@lightnvm.io, Stephen.Bates@pmcs.com, keith.busch@intel.com, =?UTF-8?q?Matias=20Bj=C3=B8rling?= Subject: [PATCH v12 2/5] gennvm: Generic NVM manager Date: Tue, 27 Oct 2015 15:55:16 +0100 Message-Id: <1445957719-5368-3-git-send-email-m@bjorling.me> X-Mailer: git-send-email 2.1.4 In-Reply-To: <1445957719-5368-1-git-send-email-m@bjorling.me> References: <1445957719-5368-1-git-send-email-m@bjorling.me> MIME-Version: 1.0 Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org X-Spam-Status: No, score=-6.8 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 The implementation for Open-Channel SSDs is divided into media management and targets. This patch implements a generic media manager for open-channel SSDs. After a media manager has been initialized, single or multiple targets can be instantiated with the media managed as the backend. Signed-off-by: Matias Bjørling --- drivers/lightnvm/Kconfig | 7 + drivers/lightnvm/Makefile | 1 + drivers/lightnvm/gennvm.c | 475 ++++++++++++++++++++++++++++++++++++++++++++++ drivers/lightnvm/gennvm.h | 46 +++++ 4 files changed, 529 insertions(+) create mode 100644 drivers/lightnvm/gennvm.c create mode 100644 drivers/lightnvm/gennvm.h diff --git a/drivers/lightnvm/Kconfig b/drivers/lightnvm/Kconfig index d4f309f..2a27971 100644 --- a/drivers/lightnvm/Kconfig +++ b/drivers/lightnvm/Kconfig @@ -25,4 +25,11 @@ config NVM_DEBUG It is required to create/remove targets without IOCTLs. +config NVM_GENNVM + tristate "Generic NVM manager for Open-Channel SSDs" + ---help--- + NVM media manager for Open-Channel SSDs that offload management + functionality to device, while keeping data placement and garbage + collection decisions on the host. + endif # NVM diff --git a/drivers/lightnvm/Makefile b/drivers/lightnvm/Makefile index 38185e9..e2428e0 100644 --- a/drivers/lightnvm/Makefile +++ b/drivers/lightnvm/Makefile @@ -3,3 +3,4 @@ # obj-$(CONFIG_NVM) := core.o +obj-$(CONFIG_NVM_GENNVM) += gennvm.o diff --git a/drivers/lightnvm/gennvm.c b/drivers/lightnvm/gennvm.c new file mode 100644 index 0000000..ba4d962 --- /dev/null +++ b/drivers/lightnvm/gennvm.c @@ -0,0 +1,475 @@ +/* + * Copyright (C) 2015 Matias Bjorling + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; see the file COPYING. If not, write to + * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, + * USA. + * + * Implementation of a generic nvm manager for Open-Channel SSDs. + */ + +#include "gennvm.h" + +static void gennvm_blocks_free(struct nvm_dev *dev) +{ + struct gen_nvm *gn = dev->mp; + struct gen_lun *lun; + int i; + + gennvm_for_each_lun(gn, lun, i) { + if (!lun->vlun.blocks) + break; + vfree(lun->vlun.blocks); + } +} + +static void gennvm_luns_free(struct nvm_dev *dev) +{ + struct gen_nvm *gn = dev->mp; + + kfree(gn->luns); +} + +static int gennvm_luns_init(struct nvm_dev *dev, struct gen_nvm *gn) +{ + struct gen_lun *lun; + int i; + + gn->luns = kcalloc(dev->nr_luns, sizeof(struct gen_lun), GFP_KERNEL); + if (!gn->luns) + return -ENOMEM; + + gennvm_for_each_lun(gn, lun, i) { + spin_lock_init(&lun->vlun.lock); + INIT_LIST_HEAD(&lun->free_list); + INIT_LIST_HEAD(&lun->used_list); + INIT_LIST_HEAD(&lun->bb_list); + + lun->reserved_blocks = 2; /* for GC only */ + lun->vlun.id = i; + lun->vlun.lun_id = i % dev->luns_per_chnl; + lun->vlun.chnl_id = i / dev->luns_per_chnl; + lun->vlun.nr_free_blocks = dev->blks_per_lun; + } + return 0; +} + +static int gennvm_block_bb(u32 lun_id, void *bb_bitmap, unsigned int nr_blocks, + void *private) +{ + struct gen_nvm *gn = private; + struct gen_lun *lun = &gn->luns[lun_id]; + struct nvm_block *block; + int i; + + if (unlikely(bitmap_empty(bb_bitmap, nr_blocks))) + return 0; + + i = -1; + while ((i = find_next_bit(bb_bitmap, nr_blocks, i + 1)) < + nr_blocks) { + block = &lun->vlun.blocks[i]; + if (!block) { + pr_err("gen_nvm: BB data is out of bounds.\n"); + return -EINVAL; + } + list_move_tail(&block->list, &lun->bb_list); + } + + return 0; +} + +static int gennvm_block_map(u64 slba, u32 nlb, __le64 *entries, void *private) +{ + struct nvm_dev *dev = private; + struct gen_nvm *gn = dev->mp; + sector_t max_pages = dev->total_pages * (dev->sec_size >> 9); + u64 elba = slba + nlb; + struct gen_lun *lun; + struct nvm_block *blk; + u64 i; + int lun_id; + + if (unlikely(elba > dev->total_pages)) { + pr_err("gen_nvm: L2P data from device is out of bounds!\n"); + return -EINVAL; + } + + for (i = 0; i < nlb; i++) { + u64 pba = le64_to_cpu(entries[i]); + + if (unlikely(pba >= max_pages && pba != U64_MAX)) { + pr_err("gen_nvm: L2P data entry is out of bounds!\n"); + return -EINVAL; + } + + /* Address zero is a special one. The first page on a disk is + * protected. It often holds internal device boot + * information. + */ + if (!pba) + continue; + + /* resolve block from physical address */ + lun_id = pba / dev->sec_per_lun; + lun = &gn->luns[lun_id]; + + /* Calculate block offset into lun */ + pba = pba - (dev->sec_per_lun * lun_id); + blk = &lun->vlun.blocks[pba / dev->sec_per_blk]; + + if (!blk->type) { + /* at this point, we don't know anything about the + * block. It's up to the FTL on top to re-etablish the + * block state + */ + list_move_tail(&blk->list, &lun->used_list); + blk->type = 1; + lun->vlun.nr_free_blocks--; + } + } + + return 0; +} + +static int gennvm_blocks_init(struct nvm_dev *dev, struct gen_nvm *gn) +{ + struct gen_lun *lun; + struct nvm_block *block; + sector_t lun_iter, blk_iter, cur_block_id = 0; + int ret; + + gennvm_for_each_lun(gn, lun, lun_iter) { + lun->vlun.blocks = vzalloc(sizeof(struct nvm_block) * + dev->blks_per_lun); + if (!lun->vlun.blocks) + return -ENOMEM; + + for (blk_iter = 0; blk_iter < dev->blks_per_lun; blk_iter++) { + block = &lun->vlun.blocks[blk_iter]; + + INIT_LIST_HEAD(&block->list); + + block->lun = &lun->vlun; + block->id = cur_block_id++; + + /* First block is reserved for device */ + if (unlikely(lun_iter == 0 && blk_iter == 0)) + continue; + + list_add_tail(&block->list, &lun->free_list); + } + + if (dev->ops->get_bb_tbl) { + ret = dev->ops->get_bb_tbl(dev->q, lun->vlun.id, + dev->blks_per_lun, gennvm_block_bb, gn); + if (ret) + pr_err("gen_nvm: could not read BB table\n"); + } + } + + if (dev->ops->get_l2p_tbl) { + ret = dev->ops->get_l2p_tbl(dev->q, 0, dev->total_pages, + gennvm_block_map, dev); + if (ret) { + pr_err("gen_nvm: could not read L2P table.\n"); + pr_warn("gen_nvm: default block initialization"); + } + } + + return 0; +} + +static int gennvm_register(struct nvm_dev *dev) +{ + struct gen_nvm *gn; + int ret; + + gn = kzalloc(sizeof(struct gen_nvm), GFP_KERNEL); + if (!gn) + return -ENOMEM; + + gn->nr_luns = dev->nr_luns; + dev->mp = gn; + + ret = gennvm_luns_init(dev, gn); + if (ret) { + pr_err("gen_nvm: could not initialize luns\n"); + goto err; + } + + ret = gennvm_blocks_init(dev, gn); + if (ret) { + pr_err("gen_nvm: could not initialize blocks\n"); + goto err; + } + + return 1; +err: + kfree(gn); + return ret; +} + +static void gennvm_unregister(struct nvm_dev *dev) +{ + gennvm_blocks_free(dev); + gennvm_luns_free(dev); + kfree(dev->mp); + dev->mp = NULL; +} + +static struct nvm_block *gennvm_get_blk(struct nvm_dev *dev, + struct nvm_lun *vlun, unsigned long flags) +{ + struct gen_lun *lun = container_of(vlun, struct gen_lun, vlun); + struct nvm_block *blk = NULL; + int is_gc = flags & NVM_IOTYPE_GC; + + BUG_ON(!lun); + + spin_lock(&vlun->lock); + + if (list_empty(&lun->free_list)) { + pr_err_ratelimited("gen_nvm: lun %u have no free pages available", + lun->vlun.id); + spin_unlock(&vlun->lock); + goto out; + } + + while (!is_gc && lun->vlun.nr_free_blocks < lun->reserved_blocks) { + spin_unlock(&vlun->lock); + goto out; + } + + blk = list_first_entry(&lun->free_list, struct nvm_block, list); + list_move_tail(&blk->list, &lun->used_list); + blk->type = 1; + + lun->vlun.nr_free_blocks--; + + spin_unlock(&vlun->lock); +out: + return blk; +} + +static void gennvm_put_blk(struct nvm_dev *dev, struct nvm_block *blk) +{ + struct nvm_lun *vlun = blk->lun; + struct gen_lun *lun = container_of(vlun, struct gen_lun, vlun); + + spin_lock(&vlun->lock); + + switch (blk->type) { + case 1: + list_move_tail(&blk->list, &lun->free_list); + lun->vlun.nr_free_blocks++; + blk->type = 0; + break; + case 2: + list_move_tail(&blk->list, &lun->bb_list); + break; + default: + BUG(); + } + + spin_unlock(&vlun->lock); +} + +static void gennvm_addr_to_generic_mode(struct nvm_dev *dev, struct nvm_rq *rqd) +{ + int i; + + if (rqd->nr_pages > 1) + for (i = 0; i < rqd->nr_pages; i++) + rqd->ppa_list[i] = addr_to_generic_mode(dev, + rqd->ppa_list[i]); + else + rqd->ppa_addr = addr_to_generic_mode(dev, rqd->ppa_addr); +} + +static void gennvm_generic_to_addr_mode(struct nvm_dev *dev, struct nvm_rq *rqd) +{ + int i; + + if (rqd->nr_pages > 1) + for (i = 0; i < rqd->nr_pages; i++) + rqd->ppa_list[i] = generic_to_addr_mode(dev, + rqd->ppa_list[i]); + else + rqd->ppa_addr = generic_to_addr_mode(dev, rqd->ppa_addr); +} + +static int gennvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd) +{ + if (!dev->ops->submit_io) + return 0; + + /* Convert address space */ + gennvm_generic_to_addr_mode(dev, rqd); + + rqd->dev = dev; + return dev->ops->submit_io(dev->q, rqd); +} + +static void gennvm_blk_set_type(struct nvm_dev *dev, struct ppa_addr *ppa, + int type) +{ + struct gen_nvm *gn = dev->mp; + struct gen_lun *lun; + struct nvm_block *blk; + + BUG_ON(ppa->g.ch > dev->nr_chnls); + BUG_ON(ppa->g.lun > dev->luns_per_chnl); + BUG_ON(ppa->g.blk > dev->blks_per_lun); + + lun = &gn->luns[ppa->g.lun * ppa->g.ch]; + blk = &lun->vlun.blocks[ppa->g.blk]; + + /* will be moved to bb list on put_blk from target */ + blk->type = type; +} + +/* mark block bad. It is expected the target recover from the error. */ +static void gennvm_mark_blk_bad(struct nvm_dev *dev, struct nvm_rq *rqd) +{ + int i; + + if (!dev->ops->set_bb) + return; + + if (dev->ops->set_bb(dev->q, rqd, 1)) + return; + + gennvm_addr_to_generic_mode(dev, rqd); + + /* look up blocks and mark them as bad */ + if (rqd->nr_pages > 1) + for (i = 0; i < rqd->nr_pages; i++) + gennvm_blk_set_type(dev, &rqd->ppa_list[i], 2); + else + gennvm_blk_set_type(dev, &rqd->ppa_addr, 2); +} + +static int gennvm_end_io(struct nvm_rq *rqd, int error) +{ + struct nvm_tgt_instance *ins = rqd->ins; + int ret = 0; + + switch (error) { + case NVM_RSP_SUCCESS: + break; + case NVM_RSP_ERR_EMPTYPAGE: + break; + case NVM_RSP_ERR_FAILWRITE: + gennvm_mark_blk_bad(rqd->dev, rqd); + default: + ret++; + } + + ret += ins->tt->end_io(rqd, error); + + return ret; +} + +static int gennvm_erase_blk(struct nvm_dev *dev, struct nvm_block *blk, + unsigned long flags) +{ + int plane_cnt = 0, pl_idx, ret; + struct ppa_addr addr; + struct nvm_rq rqd; + + if (!dev->ops->erase_block) + return 0; + + addr = block_to_ppa(dev, blk); + + if (dev->plane_mode == NVM_PLANE_SINGLE) { + rqd.nr_pages = 1; + rqd.ppa_addr = addr; + } else { + plane_cnt = (1 << dev->plane_mode); + rqd.nr_pages = plane_cnt; + + rqd.ppa_list = nvm_dev_dma_alloc(dev, GFP_KERNEL, + &rqd.dma_ppa_list); + if (!rqd.ppa_list) { + pr_err("gen_nvm: failed to allocate dma memory\n"); + return -ENOMEM; + } + + for (pl_idx = 0; pl_idx < plane_cnt; pl_idx++) { + addr.g.pl = pl_idx; + rqd.ppa_list[pl_idx] = addr; + } + } + + gennvm_generic_to_addr_mode(dev, &rqd); + + ret = dev->ops->erase_block(dev->q, &rqd); + + if (plane_cnt) + nvm_dev_dma_free(dev, rqd.ppa_list, rqd.dma_ppa_list); + + return ret; +} + +static struct nvm_lun *gennvm_get_lun(struct nvm_dev *dev, int lunid) +{ + struct gen_nvm *gn = dev->mp; + + return &gn->luns[lunid].vlun; +} + +static void gennvm_free_blocks_print(struct nvm_dev *dev) +{ + struct gen_nvm *gn = dev->mp; + struct gen_lun *lun; + unsigned int i; + + gennvm_for_each_lun(gn, lun, i) + pr_info("%s: lun%8u\t%u\n", + dev->name, i, lun->vlun.nr_free_blocks); +} + +static struct nvmm_type gennvm = { + .name = "gennvm", + .version = {0, 1, 0}, + + .register_mgr = gennvm_register, + .unregister_mgr = gennvm_unregister, + + .get_blk = gennvm_get_blk, + .put_blk = gennvm_put_blk, + + .submit_io = gennvm_submit_io, + .end_io = gennvm_end_io, + .erase_blk = gennvm_erase_blk, + + .get_lun = gennvm_get_lun, + .free_blocks_print = gennvm_free_blocks_print, +}; + +static int __init gennvm_module_init(void) +{ + return nvm_register_mgr(&gennvm); +} + +static void gennvm_module_exit(void) +{ + nvm_unregister_mgr(&gennvm); +} + +module_init(gennvm_module_init); +module_exit(gennvm_module_exit); +MODULE_LICENSE("GPL v2"); +MODULE_DESCRIPTION("Block manager for Hybrid Open-Channel SSDs"); diff --git a/drivers/lightnvm/gennvm.h b/drivers/lightnvm/gennvm.h new file mode 100644 index 0000000..d23bd35 --- /dev/null +++ b/drivers/lightnvm/gennvm.h @@ -0,0 +1,46 @@ +/* + * Copyright: Matias Bjorling + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + */ + +#ifndef GENNVM_H_ +#define GENNVM_H_ + +#include +#include + +#include + +struct gen_lun { + struct nvm_lun vlun; + + int reserved_blocks; + /* lun block lists */ + struct list_head used_list; /* In-use blocks */ + struct list_head free_list; /* Not used blocks i.e. released + * and ready for use + */ + struct list_head bb_list; /* Bad blocks. Mutually exclusive with + * free_list and used_list + */ +}; + +struct gen_nvm { + int nr_luns; + struct gen_lun *luns; +}; + +#define gennvm_for_each_lun(bm, lun, i) \ + for ((i) = 0, lun = &(bm)->luns[0]; \ + (i) < (bm)->nr_luns; (i)++, lun = &(bm)->luns[(i)]) + +#endif /* GENNVM_H_ */