[mirror_ubuntu-jammy-kernel.git] / drivers / nvme / host / lightnvm.c

/*
 * nvme-lightnvm.c - LightNVM NVMe device
 *
 * Copyright (C) 2014-2015 IT University of Copenhagen
 * Initial release: Matias Bjorling <mb@lightnvm.io>
 *
 * 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.
 *
 */

#include "nvme.h"

#ifdef CONFIG_NVM

#include <linux/nvme.h>
#include <linux/bitops.h>
#include <linux/lightnvm.h>
#include <linux/vmalloc.h>

enum nvme_nvm_admin_opcode {
	nvme_nvm_admin_identity		= 0xe2,
	nvme_nvm_admin_get_l2p_tbl	= 0xea,
	nvme_nvm_admin_get_bb_tbl	= 0xf2,
	nvme_nvm_admin_set_bb_tbl	= 0xf1,
};

struct nvme_nvm_hb_rw {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__u64			rsvd2;
	__le64			metadata;
	__le64			prp1;
	__le64			prp2;
	__le64			spba;
	__le16			length;
	__le16			control;
	__le32			dsmgmt;
	__le64			slba;
};

struct nvme_nvm_ph_rw {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__u64			rsvd2;
	__le64			metadata;
	__le64			prp1;
	__le64			prp2;
	__le64			spba;
	__le16			length;
	__le16			control;
	__le32			dsmgmt;
	__le64			resv;
};

struct nvme_nvm_identity {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__u64			rsvd[2];
	__le64			prp1;
	__le64			prp2;
	__le32			chnl_off;
	__u32			rsvd11[5];
};

struct nvme_nvm_l2ptbl {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__le32			cdw2[4];
	__le64			prp1;
	__le64			prp2;
	__le64			slba;
	__le32			nlb;
	__le16			cdw14[6];
};

struct nvme_nvm_getbbtbl {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__u64			rsvd[2];
	__le64			prp1;
	__le64			prp2;
	__le64			spba;
	__u32			rsvd4[4];
};

struct nvme_nvm_setbbtbl {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__le64			rsvd[2];
	__le64			prp1;
	__le64			prp2;
	__le64			spba;
	__le16			nlb;
	__u8			value;
	__u8			rsvd3;
	__u32			rsvd4[3];
};

struct nvme_nvm_erase_blk {
	__u8			opcode;
	__u8			flags;
	__u16			command_id;
	__le32			nsid;
	__u64			rsvd[2];
	__le64			prp1;
	__le64			prp2;
	__le64			spba;
	__le16			length;
	__le16			control;
	__le32			dsmgmt;
	__le64			resv;
};

struct nvme_nvm_command {
	union {
		struct nvme_common_command common;
		struct nvme_nvm_identity identity;
		struct nvme_nvm_hb_rw hb_rw;
		struct nvme_nvm_ph_rw ph_rw;
		struct nvme_nvm_l2ptbl l2p;
		struct nvme_nvm_getbbtbl get_bb;
		struct nvme_nvm_setbbtbl set_bb;
		struct nvme_nvm_erase_blk erase;
	};
};

struct nvme_nvm_id_group {
	__u8			mtype;
	__u8			fmtype;
	__le16			res16;
	__u8			num_ch;
	__u8			num_lun;
	__u8			num_pln;
	__u8			rsvd1;
	__le16			num_blk;
	__le16			num_pg;
	__le16			fpg_sz;
	__le16			csecs;
	__le16			sos;
	__le16			rsvd2;
	__le32			trdt;
	__le32			trdm;
	__le32			tprt;
	__le32			tprm;
	__le32			tbet;
	__le32			tbem;
	__le32			mpos;
	__le32			mccap;
	__le16			cpar;
	__u8			reserved[906];
} __packed;

struct nvme_nvm_addr_format {
	__u8			ch_offset;
	__u8			ch_len;
	__u8			lun_offset;
	__u8			lun_len;
	__u8			pln_offset;
	__u8			pln_len;
	__u8			blk_offset;
	__u8			blk_len;
	__u8			pg_offset;
	__u8			pg_len;
	__u8			sect_offset;
	__u8			sect_len;
	__u8			res[4];
} __packed;

struct nvme_nvm_id {
	__u8			ver_id;
	__u8			vmnt;
	__u8			cgrps;
	__u8			res[5];
	__le32			cap;
	__le32			dom;
	struct nvme_nvm_addr_format ppaf;
	__u8			resv[224];
	struct nvme_nvm_id_group groups[4];
} __packed;

struct nvme_nvm_bb_tbl {
	__u8	tblid[4];
	__le16	verid;
	__le16	revid;
	__le32	rvsd1;
	__le32	tblks;
	__le32	tfact;
	__le32	tgrown;
	__le32	tdresv;
	__le32	thresv;
	__le32	rsvd2[8];
	__u8	blk[0];
};

/*
 * Check we didn't inadvertently grow the command struct
 */
static inline void _nvme_nvm_check_size(void)
{
	BUILD_BUG_ON(sizeof(struct nvme_nvm_identity) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_hb_rw) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_ph_rw) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_getbbtbl) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_setbbtbl) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_l2ptbl) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_erase_blk) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_id_group) != 960);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_addr_format) != 128);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_id) != 4096);
	BUILD_BUG_ON(sizeof(struct nvme_nvm_bb_tbl) != 512);
}

static int init_grps(struct nvm_id *nvm_id, struct nvme_nvm_id *nvme_nvm_id)
{
	struct nvme_nvm_id_group *src;
	struct nvm_id_group *dst;
	int i, end;

	end = min_t(u32, 4, nvm_id->cgrps);

	for (i = 0; i < end; i++) {
		src = &nvme_nvm_id->groups[i];
		dst = &nvm_id->groups[i];

		dst->mtype = src->mtype;
		dst->fmtype = src->fmtype;
		dst->num_ch = src->num_ch;
		dst->num_lun = src->num_lun;
		dst->num_pln = src->num_pln;

		dst->num_pg = le16_to_cpu(src->num_pg);
		dst->num_blk = le16_to_cpu(src->num_blk);
		dst->fpg_sz = le16_to_cpu(src->fpg_sz);
		dst->csecs = le16_to_cpu(src->csecs);
		dst->sos = le16_to_cpu(src->sos);

		dst->trdt = le32_to_cpu(src->trdt);
		dst->trdm = le32_to_cpu(src->trdm);
		dst->tprt = le32_to_cpu(src->tprt);
		dst->tprm = le32_to_cpu(src->tprm);
		dst->tbet = le32_to_cpu(src->tbet);
		dst->tbem = le32_to_cpu(src->tbem);
		dst->mpos = le32_to_cpu(src->mpos);
		dst->mccap = le32_to_cpu(src->mccap);

		dst->cpar = le16_to_cpu(src->cpar);
	}

	return 0;
}

static int nvme_nvm_identity(struct request_queue *q, struct nvm_id *nvm_id)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_nvm_id *nvme_nvm_id;
	struct nvme_nvm_command c = {};
	int ret;

	c.identity.opcode = nvme_nvm_admin_identity;
	c.identity.nsid = cpu_to_le32(ns->ns_id);
	c.identity.chnl_off = 0;

	nvme_nvm_id = kmalloc(sizeof(struct nvme_nvm_id), GFP_KERNEL);
	if (!nvme_nvm_id)
		return -ENOMEM;

	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, nvme_nvm_id,
						sizeof(struct nvme_nvm_id));
	if (ret) {
		ret = -EIO;
		goto out;
	}

	nvm_id->ver_id = nvme_nvm_id->ver_id;
	nvm_id->vmnt = nvme_nvm_id->vmnt;
	nvm_id->cgrps = nvme_nvm_id->cgrps;
	nvm_id->cap = le32_to_cpu(nvme_nvm_id->cap);
	nvm_id->dom = le32_to_cpu(nvme_nvm_id->dom);

	ret = init_grps(nvm_id, nvme_nvm_id);
out:
	kfree(nvme_nvm_id);
	return ret;
}

static int nvme_nvm_get_l2p_tbl(struct request_queue *q, u64 slba, u32 nlb,
				nvm_l2p_update_fn *update_l2p, void *priv)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_dev *dev = ns->dev;
	struct nvme_nvm_command c = {};
	u32 len = queue_max_hw_sectors(q) << 9;
	u32 nlb_pr_rq = len / sizeof(u64);
	u64 cmd_slba = slba;
	void *entries;
	int ret = 0;

	c.l2p.opcode = nvme_nvm_admin_get_l2p_tbl;
	c.l2p.nsid = cpu_to_le32(ns->ns_id);
	entries = kmalloc(len, GFP_KERNEL);
	if (!entries)
		return -ENOMEM;

	while (nlb) {
		u32 cmd_nlb = min(nlb_pr_rq, nlb);

		c.l2p.slba = cpu_to_le64(cmd_slba);
		c.l2p.nlb = cpu_to_le32(cmd_nlb);

		ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c,
								entries, len);
		if (ret) {
			dev_err(dev->dev, "L2P table transfer failed (%d)\n",
									ret);
			ret = -EIO;
			goto out;
		}

		if (update_l2p(cmd_slba, cmd_nlb, entries, priv)) {
			ret = -EINTR;
			goto out;
		}

		cmd_slba += cmd_nlb;
		nlb -= cmd_nlb;
	}

out:
	kfree(entries);
	return ret;
}

static int nvme_nvm_get_bb_tbl(struct request_queue *q, struct ppa_addr ppa,
				int nr_blocks, nvm_bb_update_fn *update_bbtbl,
				void *priv)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_dev *dev = ns->dev;
	struct nvme_nvm_command c = {};
	struct nvme_nvm_bb_tbl *bb_tbl;
	int tblsz = sizeof(struct nvme_nvm_bb_tbl) + nr_blocks;
	int ret = 0;

	c.get_bb.opcode = nvme_nvm_admin_get_bb_tbl;
	c.get_bb.nsid = cpu_to_le32(ns->ns_id);
	c.get_bb.spba = cpu_to_le64(ppa.ppa);

	bb_tbl = kzalloc(tblsz, GFP_KERNEL);
	if (!bb_tbl)
		return -ENOMEM;

	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, bb_tbl, tblsz);
	if (ret) {
		dev_err(dev->dev, "get bad block table failed (%d)\n", ret);
		ret = -EIO;
		goto out;
	}

	if (bb_tbl->tblid[0] != 'B' || bb_tbl->tblid[1] != 'B' ||
		bb_tbl->tblid[2] != 'L' || bb_tbl->tblid[3] != 'T') {
		dev_err(dev->dev, "bbt format mismatch\n");
		ret = -EINVAL;
		goto out;
	}

	if (le16_to_cpu(bb_tbl->verid) != 1) {
		ret = -EINVAL;
		dev_err(dev->dev, "bbt version not supported\n");
		goto out;
	}

	if (le32_to_cpu(bb_tbl->tblks) != nr_blocks) {
		ret = -EINVAL;
		dev_err(dev->dev, "bbt unsuspected blocks returned (%u!=%u)",
					le32_to_cpu(bb_tbl->tblks), nr_blocks);
		goto out;
	}

	ret = update_bbtbl(ppa, nr_blocks, bb_tbl->blk, priv);
	if (ret) {
		ret = -EINTR;
		goto out;
	}

out:
	kfree(bb_tbl);
	return ret;
}

static int nvme_nvm_set_bb_tbl(struct request_queue *q, struct nvm_rq *rqd,
								int type)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_dev *dev = ns->dev;
	struct nvme_nvm_command c = {};
	int ret = 0;

	c.set_bb.opcode = nvme_nvm_admin_set_bb_tbl;
	c.set_bb.nsid = cpu_to_le32(ns->ns_id);
	c.set_bb.spba = cpu_to_le64(rqd->ppa_addr.ppa);
	c.set_bb.nlb = cpu_to_le16(rqd->nr_pages - 1);
	c.set_bb.value = type;

	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, NULL, 0);
	if (ret)
		dev_err(dev->dev, "set bad block table failed (%d)\n", ret);
	return ret;
}

static inline void nvme_nvm_rqtocmd(struct request *rq, struct nvm_rq *rqd,
				struct nvme_ns *ns, struct nvme_nvm_command *c)
{
	c->ph_rw.opcode = rqd->opcode;
	c->ph_rw.nsid = cpu_to_le32(ns->ns_id);
	c->ph_rw.spba = cpu_to_le64(rqd->ppa_addr.ppa);
	c->ph_rw.control = cpu_to_le16(rqd->flags);
	c->ph_rw.length = cpu_to_le16(rqd->nr_pages - 1);

	if (rqd->opcode == NVM_OP_HBWRITE || rqd->opcode == NVM_OP_HBREAD)
		c->hb_rw.slba = cpu_to_le64(nvme_block_nr(ns,
						rqd->bio->bi_iter.bi_sector));
}

static void nvme_nvm_end_io(struct request *rq, int error)
{
	struct nvm_rq *rqd = rq->end_io_data;
	struct nvm_dev *dev = rqd->dev;

	if (dev->mt->end_io(rqd, error))
		pr_err("nvme: err status: %x result: %lx\n",
				rq->errors, (unsigned long)rq->special);

	kfree(rq->cmd);
	blk_mq_free_request(rq);
}

static int nvme_nvm_submit_io(struct request_queue *q, struct nvm_rq *rqd)
{
	struct nvme_ns *ns = q->queuedata;
	struct request *rq;
	struct bio *bio = rqd->bio;
	struct nvme_nvm_command *cmd;

	rq = blk_mq_alloc_request(q, bio_rw(bio), GFP_KERNEL, 0);
	if (IS_ERR(rq))
		return -ENOMEM;

	cmd = kzalloc(sizeof(struct nvme_nvm_command), GFP_KERNEL);
	if (!cmd) {
		blk_mq_free_request(rq);
		return -ENOMEM;
	}

	rq->cmd_type = REQ_TYPE_DRV_PRIV;
	rq->ioprio = bio_prio(bio);

	if (bio_has_data(bio))
		rq->nr_phys_segments = bio_phys_segments(q, bio);

	rq->__data_len = bio->bi_iter.bi_size;
	rq->bio = rq->biotail = bio;

	nvme_nvm_rqtocmd(rq, rqd, ns, cmd);

	rq->cmd = (unsigned char *)cmd;
	rq->cmd_len = sizeof(struct nvme_nvm_command);
	rq->special = (void *)0;

	rq->end_io_data = rqd;

	blk_execute_rq_nowait(q, NULL, rq, 0, nvme_nvm_end_io);

	return 0;
}

static int nvme_nvm_erase_block(struct request_queue *q, struct nvm_rq *rqd)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_nvm_command c = {};

	c.erase.opcode = NVM_OP_ERASE;
	c.erase.nsid = cpu_to_le32(ns->ns_id);
	c.erase.spba = cpu_to_le64(rqd->ppa_addr.ppa);
	c.erase.length = cpu_to_le16(rqd->nr_pages - 1);

	return nvme_submit_sync_cmd(q, (struct nvme_command *)&c, NULL, 0);
}

static void *nvme_nvm_create_dma_pool(struct request_queue *q, char *name)
{
	struct nvme_ns *ns = q->queuedata;
	struct nvme_dev *dev = ns->dev;

	return dma_pool_create(name, dev->dev, PAGE_SIZE, PAGE_SIZE, 0);
}

static void nvme_nvm_destroy_dma_pool(void *pool)
{
	struct dma_pool *dma_pool = pool;

	dma_pool_destroy(dma_pool);
}

static void *nvme_nvm_dev_dma_alloc(struct request_queue *q, void *pool,
				    gfp_t mem_flags, dma_addr_t *dma_handler)
{
	return dma_pool_alloc(pool, mem_flags, dma_handler);
}

static void nvme_nvm_dev_dma_free(void *pool, void *ppa_list,
							dma_addr_t dma_handler)
{
	dma_pool_free(pool, ppa_list, dma_handler);
}

static struct nvm_dev_ops nvme_nvm_dev_ops = {
	.identity		= nvme_nvm_identity,

	.get_l2p_tbl		= nvme_nvm_get_l2p_tbl,

	.get_bb_tbl		= nvme_nvm_get_bb_tbl,
	.set_bb_tbl		= nvme_nvm_set_bb_tbl,

	.submit_io		= nvme_nvm_submit_io,
	.erase_block		= nvme_nvm_erase_block,

	.create_dma_pool	= nvme_nvm_create_dma_pool,
	.destroy_dma_pool	= nvme_nvm_destroy_dma_pool,
	.dev_dma_alloc		= nvme_nvm_dev_dma_alloc,
	.dev_dma_free		= nvme_nvm_dev_dma_free,

	.max_phys_sect		= 64,
};

int nvme_nvm_register(struct request_queue *q, char *disk_name)
{
	return nvm_register(q, disk_name, &nvme_nvm_dev_ops);
}

void nvme_nvm_unregister(struct request_queue *q, char *disk_name)
{
	nvm_unregister(disk_name);
}

int nvme_nvm_ns_supported(struct nvme_ns *ns, struct nvme_id_ns *id)
{
	struct nvme_dev *dev = ns->dev;
	struct pci_dev *pdev = to_pci_dev(dev->dev);

	/* QEMU NVMe simulator - PCI ID + Vendor specific bit */
	if (pdev->vendor == PCI_VENDOR_ID_INTEL && pdev->device == 0x5845 &&
							id->vs[0] == 0x1)
		return 1;

	/* CNEX Labs - PCI ID + Vendor specific bit */
	if (pdev->vendor == 0x1d1d && pdev->device == 0x2807 &&
							id->vs[0] == 0x1)
		return 1;

	return 0;
}
#else
int nvme_nvm_register(struct request_queue *q, char *disk_name)
{
	return 0;
}
void nvme_nvm_unregister(struct request_queue *q, char *disk_name) {};
int nvme_nvm_ns_supported(struct nvme_ns *ns, struct nvme_id_ns *id)
{
	return 0;
}
#endif /* CONFIG_NVM */
Commit	Line	Data
ca064085 MB	1	/*
	2	* nvme-lightnvm.c - LightNVM NVMe device
	3	*
	4	* Copyright (C) 2014-2015 IT University of Copenhagen
	5	* Initial release: Matias Bjorling <mb@lightnvm.io>
	6	*
	7	* This program is free software; you can redistribute it and/or
	8	* modify it under the terms of the GNU General Public License version
	9	* 2 as published by the Free Software Foundation.
	10	*
	11	* This program is distributed in the hope that it will be useful, but
	12	* WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	14	* General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with this program; see the file COPYING. If not, write to
	18	* the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139,
	19	* USA.
	20	*
	21	*/
	22
	23	#include "nvme.h"
	24
	25	#ifdef CONFIG_NVM
	26
	27	#include <linux/nvme.h>
	28	#include <linux/bitops.h>
	29	#include <linux/lightnvm.h>
	30	#include <linux/vmalloc.h>
	31
	32	enum nvme_nvm_admin_opcode {
	33	nvme_nvm_admin_identity = 0xe2,
	34	nvme_nvm_admin_get_l2p_tbl = 0xea,
	35	nvme_nvm_admin_get_bb_tbl = 0xf2,
	36	nvme_nvm_admin_set_bb_tbl = 0xf1,
	37	};
	38
	39	struct nvme_nvm_hb_rw {
	40	__u8 opcode;
	41	__u8 flags;
	42	__u16 command_id;
	43	__le32 nsid;
	44	__u64 rsvd2;
	45	__le64 metadata;
	46	__le64 prp1;
	47	__le64 prp2;
	48	__le64 spba;
	49	__le16 length;
	50	__le16 control;
	51	__le32 dsmgmt;
	52	__le64 slba;
	53	};
	54
	55	struct nvme_nvm_ph_rw {
	56	__u8 opcode;
	57	__u8 flags;
	58	__u16 command_id;
	59	__le32 nsid;
	60	__u64 rsvd2;
	61	__le64 metadata;
	62	__le64 prp1;
	63	__le64 prp2;
	64	__le64 spba;
65	__le16 length;
66	__le16 control;
67	__le32 dsmgmt;
68	__le64 resv;
69	};
70
71	struct nvme_nvm_identity {
72	__u8 opcode;
73	__u8 flags;
74	__u16 command_id;
75	__le32 nsid;
76	__u64 rsvd[2];
77	__le64 prp1;
78	__le64 prp2;
79	__le32 chnl_off;
80	__u32 rsvd11[5];
81	};
82
83	struct nvme_nvm_l2ptbl {
84	__u8 opcode;
85	__u8 flags;
86	__u16 command_id;
87	__le32 nsid;
88	__le32 cdw2[4];
89	__le64 prp1;
90	__le64 prp2;
91	__le64 slba;
92	__le32 nlb;
93	__le16 cdw14[6];
94	};
95
11450469	96	struct nvme_nvm_getbbtbl {
ca064085 MB	97	__u8 opcode;
	98	__u8 flags;
	99	__u16 command_id;
	100	__le32 nsid;
	101	__u64 rsvd[2];
	102	__le64 prp1;
	103	__le64 prp2;
11450469 MB	104	__le64 spba;
	105	__u32 rsvd4[4];
	106	};
	107
	108	struct nvme_nvm_setbbtbl {
	109	__u8 opcode;
	110	__u8 flags;
	111	__u16 command_id;
	112	__le32 nsid;
	113	__le64 rsvd[2];
	114	__le64 prp1;
	115	__le64 prp2;
	116	__le64 spba;
	117	__le16 nlb;
	118	__u8 value;
	119	__u8 rsvd3;
	120	__u32 rsvd4[3];
ca064085 MB	121	};
	122
	123	struct nvme_nvm_erase_blk {
	124	__u8 opcode;
	125	__u8 flags;
	126	__u16 command_id;
	127	__le32 nsid;
	128	__u64 rsvd[2];
	129	__le64 prp1;
	130	__le64 prp2;
	131	__le64 spba;
	132	__le16 length;
	133	__le16 control;
	134	__le32 dsmgmt;
	135	__le64 resv;
	136	};
	137
	138	struct nvme_nvm_command {
	139	union {
	140	struct nvme_common_command common;
	141	struct nvme_nvm_identity identity;
	142	struct nvme_nvm_hb_rw hb_rw;
	143	struct nvme_nvm_ph_rw ph_rw;
	144	struct nvme_nvm_l2ptbl l2p;
11450469 MB	145	struct nvme_nvm_getbbtbl get_bb;
11450469 MB	146	struct nvme_nvm_setbbtbl set_bb;
ca064085 MB	147	struct nvme_nvm_erase_blk erase;
	148	};
	149	};
	150
	151	struct nvme_nvm_id_group {
	152	__u8 mtype;
	153	__u8 fmtype;
	154	__le16 res16;
	155	__u8 num_ch;
	156	__u8 num_lun;
	157	__u8 num_pln;
36d5dbc6	158	__u8 rsvd1;
ca064085 MB	159	__le16 num_blk;
	160	__le16 num_pg;
	161	__le16 fpg_sz;
	162	__le16 csecs;
	163	__le16 sos;
36d5dbc6	164	__le16 rsvd2;
ca064085 MB	165	__le32 trdt;
	166	__le32 trdm;
	167	__le32 tprt;
	168	__le32 tprm;
	169	__le32 tbet;
	170	__le32 tbem;
	171	__le32 mpos;
12be5edf	172	__le32 mccap;
ca064085	173	__le16 cpar;
12be5edf	174	__u8 reserved[906];
ca064085 MB	175	} __packed;
	176
	177	struct nvme_nvm_addr_format {
	178	__u8 ch_offset;
	179	__u8 ch_len;
	180	__u8 lun_offset;
	181	__u8 lun_len;
	182	__u8 pln_offset;
	183	__u8 pln_len;
	184	__u8 blk_offset;
	185	__u8 blk_len;
	186	__u8 pg_offset;
	187	__u8 pg_len;
	188	__u8 sect_offset;
	189	__u8 sect_len;
	190	__u8 res[4];
	191	} __packed;
	192
	193	struct nvme_nvm_id {
	194	__u8 ver_id;
	195	__u8 vmnt;
	196	__u8 cgrps;
	197	__u8 res[5];
	198	__le32 cap;
	199	__le32 dom;
	200	struct nvme_nvm_addr_format ppaf;
7386af27	201	__u8 resv[224];
ca064085 MB	202	struct nvme_nvm_id_group groups[4];
	203	} __packed;
	204
11450469 MB	205	struct nvme_nvm_bb_tbl {
	206	__u8 tblid[4];
	207	__le16 verid;
	208	__le16 revid;
	209	__le32 rvsd1;
	210	__le32 tblks;
	211	__le32 tfact;
	212	__le32 tgrown;
	213	__le32 tdresv;
	214	__le32 thresv;
	215	__le32 rsvd2[8];
	216	__u8 blk[0];
	217	};
	218
ca064085 MB	219	/*
	220	* Check we didn't inadvertently grow the command struct
	221	*/
	222	static inline void _nvme_nvm_check_size(void)
	223	{
	224	BUILD_BUG_ON(sizeof(struct nvme_nvm_identity) != 64);
	225	BUILD_BUG_ON(sizeof(struct nvme_nvm_hb_rw) != 64);
	226	BUILD_BUG_ON(sizeof(struct nvme_nvm_ph_rw) != 64);
11450469 MB	227	BUILD_BUG_ON(sizeof(struct nvme_nvm_getbbtbl) != 64);
11450469 MB	228	BUILD_BUG_ON(sizeof(struct nvme_nvm_setbbtbl) != 64);
ca064085 MB	229	BUILD_BUG_ON(sizeof(struct nvme_nvm_l2ptbl) != 64);
	230	BUILD_BUG_ON(sizeof(struct nvme_nvm_erase_blk) != 64);
	231	BUILD_BUG_ON(sizeof(struct nvme_nvm_id_group) != 960);
	232	BUILD_BUG_ON(sizeof(struct nvme_nvm_addr_format) != 128);
	233	BUILD_BUG_ON(sizeof(struct nvme_nvm_id) != 4096);
11450469	234	BUILD_BUG_ON(sizeof(struct nvme_nvm_bb_tbl) != 512);
ca064085 MB	235	}
	236
	237	static int init_grps(struct nvm_id nvm_id, struct nvme_nvm_id nvme_nvm_id)
	238	{
	239	struct nvme_nvm_id_group *src;
	240	struct nvm_id_group *dst;
	241	int i, end;
	242
	243	end = min_t(u32, 4, nvm_id->cgrps);
	244
	245	for (i = 0; i < end; i++) {
	246	src = &nvme_nvm_id->groups[i];
	247	dst = &nvm_id->groups[i];
	248
	249	dst->mtype = src->mtype;
	250	dst->fmtype = src->fmtype;
	251	dst->num_ch = src->num_ch;
	252	dst->num_lun = src->num_lun;
	253	dst->num_pln = src->num_pln;
	254
	255	dst->num_pg = le16_to_cpu(src->num_pg);
	256	dst->num_blk = le16_to_cpu(src->num_blk);
	257	dst->fpg_sz = le16_to_cpu(src->fpg_sz);
	258	dst->csecs = le16_to_cpu(src->csecs);
	259	dst->sos = le16_to_cpu(src->sos);
	260
	261	dst->trdt = le32_to_cpu(src->trdt);
	262	dst->trdm = le32_to_cpu(src->trdm);
	263	dst->tprt = le32_to_cpu(src->tprt);
	264	dst->tprm = le32_to_cpu(src->tprm);
	265	dst->tbet = le32_to_cpu(src->tbet);
	266	dst->tbem = le32_to_cpu(src->tbem);
	267	dst->mpos = le32_to_cpu(src->mpos);
12be5edf	268	dst->mccap = le32_to_cpu(src->mccap);
ca064085 MB	269
	270	dst->cpar = le16_to_cpu(src->cpar);
	271	}
	272
	273	return 0;
	274	}
	275
	276	static int nvme_nvm_identity(struct request_queue q, struct nvm_id nvm_id)
	277	{
	278	struct nvme_ns *ns = q->queuedata;
	279	struct nvme_nvm_id *nvme_nvm_id;
	280	struct nvme_nvm_command c = {};
	281	int ret;
	282
	283	c.identity.opcode = nvme_nvm_admin_identity;
	284	c.identity.nsid = cpu_to_le32(ns->ns_id);
	285	c.identity.chnl_off = 0;
	286
	287	nvme_nvm_id = kmalloc(sizeof(struct nvme_nvm_id), GFP_KERNEL);
	288	if (!nvme_nvm_id)
	289	return -ENOMEM;
	290
	291	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, nvme_nvm_id,
	292	sizeof(struct nvme_nvm_id));
	293	if (ret) {
	294	ret = -EIO;
	295	goto out;
	296	}
	297
	298	nvm_id->ver_id = nvme_nvm_id->ver_id;
	299	nvm_id->vmnt = nvme_nvm_id->vmnt;
	300	nvm_id->cgrps = nvme_nvm_id->cgrps;
	301	nvm_id->cap = le32_to_cpu(nvme_nvm_id->cap);
	302	nvm_id->dom = le32_to_cpu(nvme_nvm_id->dom);
	303
	304	ret = init_grps(nvm_id, nvme_nvm_id);
	305	out:
	306	kfree(nvme_nvm_id);
	307	return ret;
	308	}
	309
	310	static int nvme_nvm_get_l2p_tbl(struct request_queue *q, u64 slba, u32 nlb,
	311	nvm_l2p_update_fn update_l2p, void priv)
	312	{
	313	struct nvme_ns *ns = q->queuedata;
	314	struct nvme_dev *dev = ns->dev;
	315	struct nvme_nvm_command c = {};
	316	u32 len = queue_max_hw_sectors(q) << 9;
5f436e5e	317	u32 nlb_pr_rq = len / sizeof(u64);
ca064085 MB	318	u64 cmd_slba = slba;
	319	void *entries;
	320	int ret = 0;
	321
	322	c.l2p.opcode = nvme_nvm_admin_get_l2p_tbl;
	323	c.l2p.nsid = cpu_to_le32(ns->ns_id);
	324	entries = kmalloc(len, GFP_KERNEL);
	325	if (!entries)
	326	return -ENOMEM;
	327
	328	while (nlb) {
5f436e5e	329	u32 cmd_nlb = min(nlb_pr_rq, nlb);
ca064085 MB	330
	331	c.l2p.slba = cpu_to_le64(cmd_slba);
	332	c.l2p.nlb = cpu_to_le32(cmd_nlb);
	333
	334	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c,
	335	entries, len);
	336	if (ret) {
	337	dev_err(dev->dev, "L2P table transfer failed (%d)\n",
	338	ret);
	339	ret = -EIO;
	340	goto out;
	341	}
	342
	343	if (update_l2p(cmd_slba, cmd_nlb, entries, priv)) {
	344	ret = -EINTR;
	345	goto out;
	346	}
	347
	348	cmd_slba += cmd_nlb;
	349	nlb -= cmd_nlb;
	350	}
	351
	352	out:
	353	kfree(entries);
	354	return ret;
	355	}
	356
11450469 MB	357	static int nvme_nvm_get_bb_tbl(struct request_queue *q, struct ppa_addr ppa,
	358	int nr_blocks, nvm_bb_update_fn *update_bbtbl,
	359	void *priv)
ca064085 MB	360	{
	361	struct nvme_ns *ns = q->queuedata;
	362	struct nvme_dev *dev = ns->dev;
	363	struct nvme_nvm_command c = {};
11450469 MB	364	struct nvme_nvm_bb_tbl *bb_tbl;
11450469 MB	365	int tblsz = sizeof(struct nvme_nvm_bb_tbl) + nr_blocks;
ca064085 MB	366	int ret = 0;
	367
	368	c.get_bb.opcode = nvme_nvm_admin_get_bb_tbl;
	369	c.get_bb.nsid = cpu_to_le32(ns->ns_id);
11450469	370	c.get_bb.spba = cpu_to_le64(ppa.ppa);
ca064085	371
11450469 MB	372	bb_tbl = kzalloc(tblsz, GFP_KERNEL);
	373	if (!bb_tbl)
	374	return -ENOMEM;
ca064085	375
11450469	376	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, bb_tbl, tblsz);
ca064085 MB	377	if (ret) {
	378	dev_err(dev->dev, "get bad block table failed (%d)\n", ret);
	379	ret = -EIO;
	380	goto out;
	381	}
	382
11450469 MB	383	if (bb_tbl->tblid[0] != 'B' \|\| bb_tbl->tblid[1] != 'B' \|\|
	384	bb_tbl->tblid[2] != 'L' \|\| bb_tbl->tblid[3] != 'T') {
	385	dev_err(dev->dev, "bbt format mismatch\n");
	386	ret = -EINVAL;
	387	goto out;
	388	}
	389
	390	if (le16_to_cpu(bb_tbl->verid) != 1) {
	391	ret = -EINVAL;
	392	dev_err(dev->dev, "bbt version not supported\n");
	393	goto out;
	394	}
	395
	396	if (le32_to_cpu(bb_tbl->tblks) != nr_blocks) {
	397	ret = -EINVAL;
	398	dev_err(dev->dev, "bbt unsuspected blocks returned (%u!=%u)",
	399	le32_to_cpu(bb_tbl->tblks), nr_blocks);
	400	goto out;
	401	}
	402
	403	ret = update_bbtbl(ppa, nr_blocks, bb_tbl->blk, priv);
ca064085 MB	404	if (ret) {
	405	ret = -EINTR;
	406	goto out;
	407	}
	408
	409	out:
11450469 MB	410	kfree(bb_tbl);
	411	return ret;
	412	}
	413
	414	static int nvme_nvm_set_bb_tbl(struct request_queue q, struct nvm_rq rqd,
	415	int type)
	416	{
	417	struct nvme_ns *ns = q->queuedata;
	418	struct nvme_dev *dev = ns->dev;
	419	struct nvme_nvm_command c = {};
	420	int ret = 0;
	421
	422	c.set_bb.opcode = nvme_nvm_admin_set_bb_tbl;
	423	c.set_bb.nsid = cpu_to_le32(ns->ns_id);
	424	c.set_bb.spba = cpu_to_le64(rqd->ppa_addr.ppa);
	425	c.set_bb.nlb = cpu_to_le16(rqd->nr_pages - 1);
	426	c.set_bb.value = type;
	427
	428	ret = nvme_submit_sync_cmd(q, (struct nvme_command *)&c, NULL, 0);
	429	if (ret)
	430	dev_err(dev->dev, "set bad block table failed (%d)\n", ret);
ca064085 MB	431	return ret;
	432	}
	433
	434	static inline void nvme_nvm_rqtocmd(struct request rq, struct nvm_rq rqd,
	435	struct nvme_ns ns, struct nvme_nvm_command c)
	436	{
	437	c->ph_rw.opcode = rqd->opcode;
	438	c->ph_rw.nsid = cpu_to_le32(ns->ns_id);
	439	c->ph_rw.spba = cpu_to_le64(rqd->ppa_addr.ppa);
	440	c->ph_rw.control = cpu_to_le16(rqd->flags);
	441	c->ph_rw.length = cpu_to_le16(rqd->nr_pages - 1);
	442
	443	if (rqd->opcode == NVM_OP_HBWRITE \|\| rqd->opcode == NVM_OP_HBREAD)
	444	c->hb_rw.slba = cpu_to_le64(nvme_block_nr(ns,
	445	rqd->bio->bi_iter.bi_sector));
	446	}
	447
	448	static void nvme_nvm_end_io(struct request *rq, int error)
	449	{
	450	struct nvm_rq *rqd = rq->end_io_data;
	451	struct nvm_dev *dev = rqd->dev;
	452
	453	if (dev->mt->end_io(rqd, error))
	454	pr_err("nvme: err status: %x result: %lx\n",
	455	rq->errors, (unsigned long)rq->special);
	456
	457	kfree(rq->cmd);
	458	blk_mq_free_request(rq);
	459	}
	460
	461	static int nvme_nvm_submit_io(struct request_queue q, struct nvm_rq rqd)
	462	{
	463	struct nvme_ns *ns = q->queuedata;
	464	struct request *rq;
	465	struct bio *bio = rqd->bio;
	466	struct nvme_nvm_command *cmd;
	467
	468	rq = blk_mq_alloc_request(q, bio_rw(bio), GFP_KERNEL, 0);
	469	if (IS_ERR(rq))
	470	return -ENOMEM;
	471
	472	cmd = kzalloc(sizeof(struct nvme_nvm_command), GFP_KERNEL);
	473	if (!cmd) {
	474	blk_mq_free_request(rq);
	475	return -ENOMEM;
	476	}
	477
	478	rq->cmd_type = REQ_TYPE_DRV_PRIV;
	479	rq->ioprio = bio_prio(bio);
	480
	481	if (bio_has_data(bio))
	482	rq->nr_phys_segments = bio_phys_segments(q, bio);
	483
	484	rq->__data_len = bio->bi_iter.bi_size;
	485	rq->bio = rq->biotail = bio;
	486
	487	nvme_nvm_rqtocmd(rq, rqd, ns, cmd);
	488
	489	rq->cmd = (unsigned char *)cmd;
	490	rq->cmd_len = sizeof(struct nvme_nvm_command);
	491	rq->special = (void *)0;
	492
	493	rq->end_io_data = rqd;
	494
495	blk_execute_rq_nowait(q, NULL, rq, 0, nvme_nvm_end_io);
496
497	return 0;
498	}
499
500	static int nvme_nvm_erase_block(struct request_queue q, struct nvm_rq rqd)
501	{
502	struct nvme_ns *ns = q->queuedata;
503	struct nvme_nvm_command c = {};
504
505	c.erase.opcode = NVM_OP_ERASE;
506	c.erase.nsid = cpu_to_le32(ns->ns_id);
507	c.erase.spba = cpu_to_le64(rqd->ppa_addr.ppa);
508	c.erase.length = cpu_to_le16(rqd->nr_pages - 1);
509
510	return nvme_submit_sync_cmd(q, (struct nvme_command *)&c, NULL, 0);
511	}
512
513	static void nvme_nvm_create_dma_pool(struct request_queue q, char *name)
514	{
515	struct nvme_ns *ns = q->queuedata;
516	struct nvme_dev *dev = ns->dev;
517
518	return dma_pool_create(name, dev->dev, PAGE_SIZE, PAGE_SIZE, 0);
519	}
520
521	static void nvme_nvm_destroy_dma_pool(void *pool)
522	{
523	struct dma_pool *dma_pool = pool;
524
525	dma_pool_destroy(dma_pool);
526	}
527
528	static void nvme_nvm_dev_dma_alloc(struct request_queue q, void *pool,
529	gfp_t mem_flags, dma_addr_t *dma_handler)
530	{
531	return dma_pool_alloc(pool, mem_flags, dma_handler);
532	}
533
534	static void nvme_nvm_dev_dma_free(void pool, void ppa_list,
535	dma_addr_t dma_handler)
536	{
537	dma_pool_free(pool, ppa_list, dma_handler);
538	}
539
540	static struct nvm_dev_ops nvme_nvm_dev_ops = {
541	.identity = nvme_nvm_identity,
542
543	.get_l2p_tbl = nvme_nvm_get_l2p_tbl,
544
545	.get_bb_tbl = nvme_nvm_get_bb_tbl,
11450469	546	.set_bb_tbl = nvme_nvm_set_bb_tbl,
ca064085 MB	547
	548	.submit_io = nvme_nvm_submit_io,
	549	.erase_block = nvme_nvm_erase_block,
	550
	551	.create_dma_pool = nvme_nvm_create_dma_pool,
	552	.destroy_dma_pool = nvme_nvm_destroy_dma_pool,
	553	.dev_dma_alloc = nvme_nvm_dev_dma_alloc,
	554	.dev_dma_free = nvme_nvm_dev_dma_free,
	555
	556	.max_phys_sect = 64,
	557	};
	558
	559	int nvme_nvm_register(struct request_queue q, char disk_name)
	560	{
	561	return nvm_register(q, disk_name, &nvme_nvm_dev_ops);
	562	}
	563
	564	void nvme_nvm_unregister(struct request_queue q, char disk_name)
	565	{
	566	nvm_unregister(disk_name);
	567	}
	568
	569	int nvme_nvm_ns_supported(struct nvme_ns ns, struct nvme_id_ns id)
	570	{
	571	struct nvme_dev *dev = ns->dev;
	572	struct pci_dev *pdev = to_pci_dev(dev->dev);
	573
	574	/* QEMU NVMe simulator - PCI ID + Vendor specific bit */
	575	if (pdev->vendor == PCI_VENDOR_ID_INTEL && pdev->device == 0x5845 &&
	576	id->vs[0] == 0x1)
	577	return 1;
	578
	579	/* CNEX Labs - PCI ID + Vendor specific bit */
	580	if (pdev->vendor == 0x1d1d && pdev->device == 0x2807 &&
	581	id->vs[0] == 0x1)
	582	return 1;
	583
	584	return 0;
	585	}
	586	#else
	587	int nvme_nvm_register(struct request_queue q, char disk_name)
	588	{
	589	return 0;
	590	}
	591	void nvme_nvm_unregister(struct request_queue q, char disk_name) {};
	592	int nvme_nvm_ns_supported(struct nvme_ns ns, struct nvme_id_ns id)
	593	{
	594	return 0;
	595	}
	596	#endif /* CONFIG_NVM */