* [GIT PULL 01/25] null_blk: remove lightnvm support
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
@ 2018-01-05 13:15 ` Matias Bjørling
2018-01-05 13:15 ` [GIT PULL 02/25] lightnvm: remove rrpc Matias Bjørling
` (24 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:15 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
With rrpc to be removed, the null_blk lightnvm support is no longer
functional. Remove the lightnvm implementation and maybe add it to
another module in the future if someone takes on the challenge.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/block/null_blk.c | 220 +----------------------------------------------
1 file changed, 3 insertions(+), 217 deletions(-)
diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c
index ccb9975..1e1981c 100644
--- a/drivers/block/null_blk.c
+++ b/drivers/block/null_blk.c
@@ -12,7 +12,6 @@
#include <linux/slab.h>
#include <linux/blk-mq.h>
#include <linux/hrtimer.h>
-#include <linux/lightnvm.h>
#include <linux/configfs.h>
#include <linux/badblocks.h>
@@ -107,7 +106,6 @@ struct nullb_device {
unsigned int hw_queue_depth; /* queue depth */
unsigned int index; /* index of the disk, only valid with a disk */
unsigned int mbps; /* Bandwidth throttle cap (in MB/s) */
- bool use_lightnvm; /* register as a LightNVM device */
bool blocking; /* blocking blk-mq device */
bool use_per_node_hctx; /* use per-node allocation for hardware context */
bool power; /* power on/off the device */
@@ -121,7 +119,6 @@ struct nullb {
unsigned int index;
struct request_queue *q;
struct gendisk *disk;
- struct nvm_dev *ndev;
struct blk_mq_tag_set *tag_set;
struct blk_mq_tag_set __tag_set;
unsigned int queue_depth;
@@ -139,7 +136,6 @@ static LIST_HEAD(nullb_list);
static struct mutex lock;
static int null_major;
static DEFINE_IDA(nullb_indexes);
-static struct kmem_cache *ppa_cache;
static struct blk_mq_tag_set tag_set;
enum {
@@ -208,10 +204,6 @@ static int nr_devices = 1;
module_param(nr_devices, int, S_IRUGO);
MODULE_PARM_DESC(nr_devices, "Number of devices to register");
-static bool g_use_lightnvm;
-module_param_named(use_lightnvm, g_use_lightnvm, bool, S_IRUGO);
-MODULE_PARM_DESC(use_lightnvm, "Register as a LightNVM device");
-
static bool g_blocking;
module_param_named(blocking, g_blocking, bool, S_IRUGO);
MODULE_PARM_DESC(blocking, "Register as a blocking blk-mq driver device");
@@ -345,7 +337,6 @@ NULLB_DEVICE_ATTR(blocksize, uint);
NULLB_DEVICE_ATTR(irqmode, uint);
NULLB_DEVICE_ATTR(hw_queue_depth, uint);
NULLB_DEVICE_ATTR(index, uint);
-NULLB_DEVICE_ATTR(use_lightnvm, bool);
NULLB_DEVICE_ATTR(blocking, bool);
NULLB_DEVICE_ATTR(use_per_node_hctx, bool);
NULLB_DEVICE_ATTR(memory_backed, bool);
@@ -455,7 +446,6 @@ static struct configfs_attribute *nullb_device_attrs[] = {
&nullb_device_attr_irqmode,
&nullb_device_attr_hw_queue_depth,
&nullb_device_attr_index,
- &nullb_device_attr_use_lightnvm,
&nullb_device_attr_blocking,
&nullb_device_attr_use_per_node_hctx,
&nullb_device_attr_power,
@@ -573,7 +563,6 @@ static struct nullb_device *null_alloc_dev(void)
dev->blocksize = g_bs;
dev->irqmode = g_irqmode;
dev->hw_queue_depth = g_hw_queue_depth;
- dev->use_lightnvm = g_use_lightnvm;
dev->blocking = g_blocking;
dev->use_per_node_hctx = g_use_per_node_hctx;
return dev;
@@ -1423,170 +1412,6 @@ static void cleanup_queues(struct nullb *nullb)
kfree(nullb->queues);
}
-#ifdef CONFIG_NVM
-
-static void null_lnvm_end_io(struct request *rq, blk_status_t status)
-{
- struct nvm_rq *rqd = rq->end_io_data;
-
- /* XXX: lighnvm core seems to expect NVM_RSP_* values here.. */
- rqd->error = status ? -EIO : 0;
- nvm_end_io(rqd);
-
- blk_put_request(rq);
-}
-
-static int null_lnvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
-{
- struct request_queue *q = dev->q;
- struct request *rq;
- struct bio *bio = rqd->bio;
-
- rq = blk_mq_alloc_request(q,
- op_is_write(bio_op(bio)) ? REQ_OP_DRV_OUT : REQ_OP_DRV_IN, 0);
- if (IS_ERR(rq))
- return -ENOMEM;
-
- blk_init_request_from_bio(rq, bio);
-
- rq->end_io_data = rqd;
-
- blk_execute_rq_nowait(q, NULL, rq, 0, null_lnvm_end_io);
-
- return 0;
-}
-
-static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
-{
- struct nullb *nullb = dev->q->queuedata;
- sector_t size = (sector_t)nullb->dev->size * 1024 * 1024ULL;
- sector_t blksize;
- struct nvm_id_group *grp;
-
- id->ver_id = 0x1;
- id->vmnt = 0;
- id->cap = 0x2;
- id->dom = 0x1;
-
- id->ppaf.blk_offset = 0;
- id->ppaf.blk_len = 16;
- id->ppaf.pg_offset = 16;
- id->ppaf.pg_len = 16;
- id->ppaf.sect_offset = 32;
- id->ppaf.sect_len = 8;
- id->ppaf.pln_offset = 40;
- id->ppaf.pln_len = 8;
- id->ppaf.lun_offset = 48;
- id->ppaf.lun_len = 8;
- id->ppaf.ch_offset = 56;
- id->ppaf.ch_len = 8;
-
- sector_div(size, nullb->dev->blocksize); /* convert size to pages */
- size >>= 8; /* concert size to pgs pr blk */
- grp = &id->grp;
- grp->mtype = 0;
- grp->fmtype = 0;
- grp->num_ch = 1;
- grp->num_pg = 256;
- blksize = size;
- size >>= 16;
- grp->num_lun = size + 1;
- sector_div(blksize, grp->num_lun);
- grp->num_blk = blksize;
- grp->num_pln = 1;
-
- grp->fpg_sz = nullb->dev->blocksize;
- grp->csecs = nullb->dev->blocksize;
- grp->trdt = 25000;
- grp->trdm = 25000;
- grp->tprt = 500000;
- grp->tprm = 500000;
- grp->tbet = 1500000;
- grp->tbem = 1500000;
- grp->mpos = 0x010101; /* single plane rwe */
- grp->cpar = nullb->dev->hw_queue_depth;
-
- return 0;
-}
-
-static void *null_lnvm_create_dma_pool(struct nvm_dev *dev, char *name)
-{
- mempool_t *virtmem_pool;
-
- virtmem_pool = mempool_create_slab_pool(64, ppa_cache);
- if (!virtmem_pool) {
- pr_err("null_blk: Unable to create virtual memory pool\n");
- return NULL;
- }
-
- return virtmem_pool;
-}
-
-static void null_lnvm_destroy_dma_pool(void *pool)
-{
- mempool_destroy(pool);
-}
-
-static void *null_lnvm_dev_dma_alloc(struct nvm_dev *dev, void *pool,
- gfp_t mem_flags, dma_addr_t *dma_handler)
-{
- return mempool_alloc(pool, mem_flags);
-}
-
-static void null_lnvm_dev_dma_free(void *pool, void *entry,
- dma_addr_t dma_handler)
-{
- mempool_free(entry, pool);
-}
-
-static struct nvm_dev_ops null_lnvm_dev_ops = {
- .identity = null_lnvm_id,
- .submit_io = null_lnvm_submit_io,
-
- .create_dma_pool = null_lnvm_create_dma_pool,
- .destroy_dma_pool = null_lnvm_destroy_dma_pool,
- .dev_dma_alloc = null_lnvm_dev_dma_alloc,
- .dev_dma_free = null_lnvm_dev_dma_free,
-
- /* Simulate nvme protocol restriction */
- .max_phys_sect = 64,
-};
-
-static int null_nvm_register(struct nullb *nullb)
-{
- struct nvm_dev *dev;
- int rv;
-
- dev = nvm_alloc_dev(0);
- if (!dev)
- return -ENOMEM;
-
- dev->q = nullb->q;
- memcpy(dev->name, nullb->disk_name, DISK_NAME_LEN);
- dev->ops = &null_lnvm_dev_ops;
-
- rv = nvm_register(dev);
- if (rv) {
- kfree(dev);
- return rv;
- }
- nullb->ndev = dev;
- return 0;
-}
-
-static void null_nvm_unregister(struct nullb *nullb)
-{
- nvm_unregister(nullb->ndev);
-}
-#else
-static int null_nvm_register(struct nullb *nullb)
-{
- pr_err("null_blk: CONFIG_NVM needs to be enabled for LightNVM\n");
- return -EINVAL;
-}
-static void null_nvm_unregister(struct nullb *nullb) {}
-#endif /* CONFIG_NVM */
-
static void null_del_dev(struct nullb *nullb)
{
struct nullb_device *dev = nullb->dev;
@@ -1595,10 +1420,7 @@ static void null_del_dev(struct nullb *nullb)
list_del_init(&nullb->list);
- if (dev->use_lightnvm)
- null_nvm_unregister(nullb);
- else
- del_gendisk(nullb->disk);
+ del_gendisk(nullb->disk);
if (test_bit(NULLB_DEV_FL_THROTTLED, &nullb->dev->flags)) {
hrtimer_cancel(&nullb->bw_timer);
@@ -1610,8 +1432,7 @@ static void null_del_dev(struct nullb *nullb)
if (dev->queue_mode == NULL_Q_MQ &&
nullb->tag_set == &nullb->__tag_set)
blk_mq_free_tag_set(nullb->tag_set);
- if (!dev->use_lightnvm)
- put_disk(nullb->disk);
+ put_disk(nullb->disk);
cleanup_queues(nullb);
if (null_cache_active(nullb))
null_free_device_storage(nullb->dev, true);
@@ -1775,11 +1596,6 @@ static void null_validate_conf(struct nullb_device *dev)
{
dev->blocksize = round_down(dev->blocksize, 512);
dev->blocksize = clamp_t(unsigned int, dev->blocksize, 512, 4096);
- if (dev->use_lightnvm && dev->blocksize != 4096)
- dev->blocksize = 4096;
-
- if (dev->use_lightnvm && dev->queue_mode != NULL_Q_MQ)
- dev->queue_mode = NULL_Q_MQ;
if (dev->queue_mode == NULL_Q_MQ && dev->use_per_node_hctx) {
if (dev->submit_queues != nr_online_nodes)
@@ -1895,11 +1711,7 @@ static int null_add_dev(struct nullb_device *dev)
sprintf(nullb->disk_name, "nullb%d", nullb->index);
- if (dev->use_lightnvm)
- rv = null_nvm_register(nullb);
- else
- rv = null_gendisk_register(nullb);
-
+ rv = null_gendisk_register(nullb);
if (rv)
goto out_cleanup_blk_queue;
@@ -1938,18 +1750,6 @@ static int __init null_init(void)
g_bs = PAGE_SIZE;
}
- if (g_use_lightnvm && g_bs != 4096) {
- pr_warn("null_blk: LightNVM only supports 4k block size\n");
- pr_warn("null_blk: defaults block size to 4k\n");
- g_bs = 4096;
- }
-
- if (g_use_lightnvm && g_queue_mode != NULL_Q_MQ) {
- pr_warn("null_blk: LightNVM only supported for blk-mq\n");
- pr_warn("null_blk: defaults queue mode to blk-mq\n");
- g_queue_mode = NULL_Q_MQ;
- }
-
if (g_queue_mode == NULL_Q_MQ && g_use_per_node_hctx) {
if (g_submit_queues != nr_online_nodes) {
pr_warn("null_blk: submit_queues param is set to %u.\n",
@@ -1982,16 +1782,6 @@ static int __init null_init(void)
goto err_conf;
}
- if (g_use_lightnvm) {
- ppa_cache = kmem_cache_create("ppa_cache", 64 * sizeof(u64),
- 0, 0, NULL);
- if (!ppa_cache) {
- pr_err("null_blk: unable to create ppa cache\n");
- ret = -ENOMEM;
- goto err_ppa;
- }
- }
-
for (i = 0; i < nr_devices; i++) {
dev = null_alloc_dev();
if (!dev) {
@@ -2015,8 +1805,6 @@ static int __init null_init(void)
null_del_dev(nullb);
null_free_dev(dev);
}
- kmem_cache_destroy(ppa_cache);
-err_ppa:
unregister_blkdev(null_major, "nullb");
err_conf:
configfs_unregister_subsystem(&nullb_subsys);
@@ -2047,8 +1835,6 @@ static void __exit null_exit(void)
if (g_queue_mode == NULL_Q_MQ && shared_tags)
blk_mq_free_tag_set(&tag_set);
-
- kmem_cache_destroy(ppa_cache);
}
module_init(null_init);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 02/25] lightnvm: remove rrpc
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
2018-01-05 13:15 ` [GIT PULL 01/25] null_blk: remove lightnvm support Matias Bjørling
@ 2018-01-05 13:15 ` Matias Bjørling
2018-01-05 13:15 ` [GIT PULL 03/25] lightnvm: use internal pblk methods Matias Bjørling
` (23 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:15 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
The hybrid mode for 1.2 revision was deprecated, and have
no users. Remove to make it easier to move to the 2.0 revision.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/Kconfig | 7 -
drivers/lightnvm/Makefile | 1 -
drivers/lightnvm/rrpc.c | 1625 ---------------------------------------------
drivers/lightnvm/rrpc.h | 290 --------
4 files changed, 1923 deletions(-)
delete mode 100644 drivers/lightnvm/rrpc.c
delete mode 100644 drivers/lightnvm/rrpc.h
diff --git a/drivers/lightnvm/Kconfig b/drivers/lightnvm/Kconfig
index 2a953ef..10c0898 100644
--- a/drivers/lightnvm/Kconfig
+++ b/drivers/lightnvm/Kconfig
@@ -27,13 +27,6 @@ config NVM_DEBUG
It is required to create/remove targets without IOCTLs.
-config NVM_RRPC
- tristate "Round-robin Hybrid Open-Channel SSD target"
- ---help---
- Allows an open-channel SSD to be exposed as a block device to the
- host. The target is implemented using a linear mapping table and
- cost-based garbage collection. It is optimized for 4K IO sizes.
-
config NVM_PBLK
tristate "Physical Block Device Open-Channel SSD target"
---help---
diff --git a/drivers/lightnvm/Makefile b/drivers/lightnvm/Makefile
index 2c3fd9d..97d9d7c 100644
--- a/drivers/lightnvm/Makefile
+++ b/drivers/lightnvm/Makefile
@@ -4,7 +4,6 @@
#
obj-$(CONFIG_NVM) := core.o
-obj-$(CONFIG_NVM_RRPC) += rrpc.o
obj-$(CONFIG_NVM_PBLK) += pblk.o
pblk-y := pblk-init.o pblk-core.o pblk-rb.o \
pblk-write.o pblk-cache.o pblk-read.o \
diff --git a/drivers/lightnvm/rrpc.c b/drivers/lightnvm/rrpc.c
deleted file mode 100644
index 0993c14..0000000
--- a/drivers/lightnvm/rrpc.c
+++ /dev/null
@@ -1,1625 +0,0 @@
-/*
- * Copyright (C) 2015 IT University of Copenhagen
- * Initial release: Matias Bjorling <m@bjorling.me>
- *
- * 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.
- *
- * Implementation of a Round-robin page-based Hybrid FTL for Open-channel SSDs.
- */
-
-#include "rrpc.h"
-
-static struct kmem_cache *rrpc_gcb_cache, *rrpc_rq_cache;
-static DECLARE_RWSEM(rrpc_lock);
-
-static int rrpc_submit_io(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags);
-
-#define rrpc_for_each_lun(rrpc, rlun, i) \
- for ((i) = 0, rlun = &(rrpc)->luns[0]; \
- (i) < (rrpc)->nr_luns; (i)++, rlun = &(rrpc)->luns[(i)])
-
-static void rrpc_page_invalidate(struct rrpc *rrpc, struct rrpc_addr *a)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_block *rblk = a->rblk;
- unsigned int pg_offset;
-
- lockdep_assert_held(&rrpc->rev_lock);
-
- if (a->addr == ADDR_EMPTY || !rblk)
- return;
-
- spin_lock(&rblk->lock);
-
- div_u64_rem(a->addr, dev->geo.sec_per_blk, &pg_offset);
- WARN_ON(test_and_set_bit(pg_offset, rblk->invalid_pages));
- rblk->nr_invalid_pages++;
-
- spin_unlock(&rblk->lock);
-
- rrpc->rev_trans_map[a->addr].addr = ADDR_EMPTY;
-}
-
-static void rrpc_invalidate_range(struct rrpc *rrpc, sector_t slba,
- unsigned int len)
-{
- sector_t i;
-
- spin_lock(&rrpc->rev_lock);
- for (i = slba; i < slba + len; i++) {
- struct rrpc_addr *gp = &rrpc->trans_map[i];
-
- rrpc_page_invalidate(rrpc, gp);
- gp->rblk = NULL;
- }
- spin_unlock(&rrpc->rev_lock);
-}
-
-static struct nvm_rq *rrpc_inflight_laddr_acquire(struct rrpc *rrpc,
- sector_t laddr, unsigned int pages)
-{
- struct nvm_rq *rqd;
- struct rrpc_inflight_rq *inf;
-
- rqd = mempool_alloc(rrpc->rq_pool, GFP_ATOMIC);
- if (!rqd)
- return ERR_PTR(-ENOMEM);
-
- inf = rrpc_get_inflight_rq(rqd);
- if (rrpc_lock_laddr(rrpc, laddr, pages, inf)) {
- mempool_free(rqd, rrpc->rq_pool);
- return NULL;
- }
-
- return rqd;
-}
-
-static void rrpc_inflight_laddr_release(struct rrpc *rrpc, struct nvm_rq *rqd)
-{
- struct rrpc_inflight_rq *inf = rrpc_get_inflight_rq(rqd);
-
- rrpc_unlock_laddr(rrpc, inf);
-
- mempool_free(rqd, rrpc->rq_pool);
-}
-
-static void rrpc_discard(struct rrpc *rrpc, struct bio *bio)
-{
- sector_t slba = bio->bi_iter.bi_sector / NR_PHY_IN_LOG;
- sector_t len = bio->bi_iter.bi_size / RRPC_EXPOSED_PAGE_SIZE;
- struct nvm_rq *rqd;
-
- while (1) {
- rqd = rrpc_inflight_laddr_acquire(rrpc, slba, len);
- if (rqd)
- break;
-
- schedule();
- }
-
- if (IS_ERR(rqd)) {
- pr_err("rrpc: unable to acquire inflight IO\n");
- bio_io_error(bio);
- return;
- }
-
- rrpc_invalidate_range(rrpc, slba, len);
- rrpc_inflight_laddr_release(rrpc, rqd);
-}
-
-static int block_is_full(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
-
- return (rblk->next_page == dev->geo.sec_per_blk);
-}
-
-/* Calculate relative addr for the given block, considering instantiated LUNs */
-static u64 block_to_rel_addr(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_lun *rlun = rblk->rlun;
-
- return rlun->id * dev->geo.sec_per_blk;
-}
-
-static struct ppa_addr rrpc_ppa_to_gaddr(struct nvm_tgt_dev *dev,
- struct rrpc_addr *gp)
-{
- struct rrpc_block *rblk = gp->rblk;
- struct rrpc_lun *rlun = rblk->rlun;
- u64 addr = gp->addr;
- struct ppa_addr paddr;
-
- paddr.ppa = addr;
- paddr = rrpc_linear_to_generic_addr(&dev->geo, paddr);
- paddr.g.ch = rlun->bppa.g.ch;
- paddr.g.lun = rlun->bppa.g.lun;
- paddr.g.blk = rblk->id;
-
- return paddr;
-}
-
-/* requires lun->lock taken */
-static void rrpc_set_lun_cur(struct rrpc_lun *rlun, struct rrpc_block *new_rblk,
- struct rrpc_block **cur_rblk)
-{
- struct rrpc *rrpc = rlun->rrpc;
-
- if (*cur_rblk) {
- spin_lock(&(*cur_rblk)->lock);
- WARN_ON(!block_is_full(rrpc, *cur_rblk));
- spin_unlock(&(*cur_rblk)->lock);
- }
- *cur_rblk = new_rblk;
-}
-
-static struct rrpc_block *__rrpc_get_blk(struct rrpc *rrpc,
- struct rrpc_lun *rlun)
-{
- struct rrpc_block *rblk = NULL;
-
- if (list_empty(&rlun->free_list))
- goto out;
-
- rblk = list_first_entry(&rlun->free_list, struct rrpc_block, list);
-
- list_move_tail(&rblk->list, &rlun->used_list);
- rblk->state = NVM_BLK_ST_TGT;
- rlun->nr_free_blocks--;
-
-out:
- return rblk;
-}
-
-static struct rrpc_block *rrpc_get_blk(struct rrpc *rrpc, struct rrpc_lun *rlun,
- unsigned long flags)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_block *rblk;
- int is_gc = flags & NVM_IOTYPE_GC;
-
- spin_lock(&rlun->lock);
- if (!is_gc && rlun->nr_free_blocks < rlun->reserved_blocks) {
- pr_err("nvm: rrpc: cannot give block to non GC request\n");
- spin_unlock(&rlun->lock);
- return NULL;
- }
-
- rblk = __rrpc_get_blk(rrpc, rlun);
- if (!rblk) {
- pr_err("nvm: rrpc: cannot get new block\n");
- spin_unlock(&rlun->lock);
- return NULL;
- }
- spin_unlock(&rlun->lock);
-
- bitmap_zero(rblk->invalid_pages, dev->geo.sec_per_blk);
- rblk->next_page = 0;
- rblk->nr_invalid_pages = 0;
- atomic_set(&rblk->data_cmnt_size, 0);
-
- return rblk;
-}
-
-static void rrpc_put_blk(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct rrpc_lun *rlun = rblk->rlun;
-
- spin_lock(&rlun->lock);
- if (rblk->state & NVM_BLK_ST_TGT) {
- list_move_tail(&rblk->list, &rlun->free_list);
- rlun->nr_free_blocks++;
- rblk->state = NVM_BLK_ST_FREE;
- } else if (rblk->state & NVM_BLK_ST_BAD) {
- list_move_tail(&rblk->list, &rlun->bb_list);
- rblk->state = NVM_BLK_ST_BAD;
- } else {
- WARN_ON_ONCE(1);
- pr_err("rrpc: erroneous type (ch:%d,lun:%d,blk%d-> %u)\n",
- rlun->bppa.g.ch, rlun->bppa.g.lun,
- rblk->id, rblk->state);
- list_move_tail(&rblk->list, &rlun->bb_list);
- }
- spin_unlock(&rlun->lock);
-}
-
-static void rrpc_put_blks(struct rrpc *rrpc)
-{
- struct rrpc_lun *rlun;
- int i;
-
- for (i = 0; i < rrpc->nr_luns; i++) {
- rlun = &rrpc->luns[i];
- if (rlun->cur)
- rrpc_put_blk(rrpc, rlun->cur);
- if (rlun->gc_cur)
- rrpc_put_blk(rrpc, rlun->gc_cur);
- }
-}
-
-static struct rrpc_lun *get_next_lun(struct rrpc *rrpc)
-{
- int next = atomic_inc_return(&rrpc->next_lun);
-
- return &rrpc->luns[next % rrpc->nr_luns];
-}
-
-static void rrpc_gc_kick(struct rrpc *rrpc)
-{
- struct rrpc_lun *rlun;
- unsigned int i;
-
- for (i = 0; i < rrpc->nr_luns; i++) {
- rlun = &rrpc->luns[i];
- queue_work(rrpc->krqd_wq, &rlun->ws_gc);
- }
-}
-
-/*
- * timed GC every interval.
- */
-static void rrpc_gc_timer(struct timer_list *t)
-{
- struct rrpc *rrpc = from_timer(rrpc, t, gc_timer);
-
- rrpc_gc_kick(rrpc);
- mod_timer(&rrpc->gc_timer, jiffies + msecs_to_jiffies(10));
-}
-
-static void rrpc_end_sync_bio(struct bio *bio)
-{
- struct completion *waiting = bio->bi_private;
-
- if (bio->bi_status)
- pr_err("nvm: gc request failed (%u).\n", bio->bi_status);
-
- complete(waiting);
-}
-
-/*
- * rrpc_move_valid_pages -- migrate live data off the block
- * @rrpc: the 'rrpc' structure
- * @block: the block from which to migrate live pages
- *
- * Description:
- * GC algorithms may call this function to migrate remaining live
- * pages off the block prior to erasing it. This function blocks
- * further execution until the operation is complete.
- */
-static int rrpc_move_valid_pages(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct request_queue *q = dev->q;
- struct rrpc_rev_addr *rev;
- struct nvm_rq *rqd;
- struct bio *bio;
- struct page *page;
- int slot;
- int nr_sec_per_blk = dev->geo.sec_per_blk;
- u64 phys_addr;
- DECLARE_COMPLETION_ONSTACK(wait);
-
- if (bitmap_full(rblk->invalid_pages, nr_sec_per_blk))
- return 0;
-
- bio = bio_alloc(GFP_NOIO, 1);
- if (!bio) {
- pr_err("nvm: could not alloc bio to gc\n");
- return -ENOMEM;
- }
-
- page = mempool_alloc(rrpc->page_pool, GFP_NOIO);
-
- while ((slot = find_first_zero_bit(rblk->invalid_pages,
- nr_sec_per_blk)) < nr_sec_per_blk) {
-
- /* Lock laddr */
- phys_addr = rrpc_blk_to_ppa(rrpc, rblk) + slot;
-
-try:
- spin_lock(&rrpc->rev_lock);
- /* Get logical address from physical to logical table */
- rev = &rrpc->rev_trans_map[phys_addr];
- /* already updated by previous regular write */
- if (rev->addr == ADDR_EMPTY) {
- spin_unlock(&rrpc->rev_lock);
- continue;
- }
-
- rqd = rrpc_inflight_laddr_acquire(rrpc, rev->addr, 1);
- if (IS_ERR_OR_NULL(rqd)) {
- spin_unlock(&rrpc->rev_lock);
- schedule();
- goto try;
- }
-
- spin_unlock(&rrpc->rev_lock);
-
- /* Perform read to do GC */
- bio->bi_iter.bi_sector = rrpc_get_sector(rev->addr);
- bio_set_op_attrs(bio, REQ_OP_READ, 0);
- bio->bi_private = &wait;
- bio->bi_end_io = rrpc_end_sync_bio;
-
- /* TODO: may fail when EXP_PG_SIZE > PAGE_SIZE */
- bio_add_pc_page(q, bio, page, RRPC_EXPOSED_PAGE_SIZE, 0);
-
- if (rrpc_submit_io(rrpc, bio, rqd, NVM_IOTYPE_GC)) {
- pr_err("rrpc: gc read failed.\n");
- rrpc_inflight_laddr_release(rrpc, rqd);
- goto finished;
- }
- wait_for_completion_io(&wait);
- if (bio->bi_status) {
- rrpc_inflight_laddr_release(rrpc, rqd);
- goto finished;
- }
-
- bio_reset(bio);
- reinit_completion(&wait);
-
- bio->bi_iter.bi_sector = rrpc_get_sector(rev->addr);
- bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
- bio->bi_private = &wait;
- bio->bi_end_io = rrpc_end_sync_bio;
-
- bio_add_pc_page(q, bio, page, RRPC_EXPOSED_PAGE_SIZE, 0);
-
- /* turn the command around and write the data back to a new
- * address
- */
- if (rrpc_submit_io(rrpc, bio, rqd, NVM_IOTYPE_GC)) {
- pr_err("rrpc: gc write failed.\n");
- rrpc_inflight_laddr_release(rrpc, rqd);
- goto finished;
- }
- wait_for_completion_io(&wait);
-
- rrpc_inflight_laddr_release(rrpc, rqd);
- if (bio->bi_status)
- goto finished;
-
- bio_reset(bio);
- }
-
-finished:
- mempool_free(page, rrpc->page_pool);
- bio_put(bio);
-
- if (!bitmap_full(rblk->invalid_pages, nr_sec_per_blk)) {
- pr_err("nvm: failed to garbage collect block\n");
- return -EIO;
- }
-
- return 0;
-}
-
-static void rrpc_block_gc(struct work_struct *work)
-{
- struct rrpc_block_gc *gcb = container_of(work, struct rrpc_block_gc,
- ws_gc);
- struct rrpc *rrpc = gcb->rrpc;
- struct rrpc_block *rblk = gcb->rblk;
- struct rrpc_lun *rlun = rblk->rlun;
- struct ppa_addr ppa;
-
- mempool_free(gcb, rrpc->gcb_pool);
- pr_debug("nvm: block 'ch:%d,lun:%d,blk:%d' being reclaimed\n",
- rlun->bppa.g.ch, rlun->bppa.g.lun,
- rblk->id);
-
- if (rrpc_move_valid_pages(rrpc, rblk))
- goto put_back;
-
- ppa.ppa = 0;
- ppa.g.ch = rlun->bppa.g.ch;
- ppa.g.lun = rlun->bppa.g.lun;
- ppa.g.blk = rblk->id;
-
- if (nvm_erase_sync(rrpc->dev, &ppa, 1))
- goto put_back;
-
- rrpc_put_blk(rrpc, rblk);
-
- return;
-
-put_back:
- spin_lock(&rlun->lock);
- list_add_tail(&rblk->prio, &rlun->prio_list);
- spin_unlock(&rlun->lock);
-}
-
-/* the block with highest number of invalid pages, will be in the beginning
- * of the list
- */
-static struct rrpc_block *rblk_max_invalid(struct rrpc_block *ra,
- struct rrpc_block *rb)
-{
- if (ra->nr_invalid_pages == rb->nr_invalid_pages)
- return ra;
-
- return (ra->nr_invalid_pages < rb->nr_invalid_pages) ? rb : ra;
-}
-
-/* linearly find the block with highest number of invalid pages
- * requires lun->lock
- */
-static struct rrpc_block *block_prio_find_max(struct rrpc_lun *rlun)
-{
- struct list_head *prio_list = &rlun->prio_list;
- struct rrpc_block *rblk, *max;
-
- BUG_ON(list_empty(prio_list));
-
- max = list_first_entry(prio_list, struct rrpc_block, prio);
- list_for_each_entry(rblk, prio_list, prio)
- max = rblk_max_invalid(max, rblk);
-
- return max;
-}
-
-static void rrpc_lun_gc(struct work_struct *work)
-{
- struct rrpc_lun *rlun = container_of(work, struct rrpc_lun, ws_gc);
- struct rrpc *rrpc = rlun->rrpc;
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_block_gc *gcb;
- unsigned int nr_blocks_need;
-
- nr_blocks_need = dev->geo.blks_per_lun / GC_LIMIT_INVERSE;
-
- if (nr_blocks_need < rrpc->nr_luns)
- nr_blocks_need = rrpc->nr_luns;
-
- spin_lock(&rlun->lock);
- while (nr_blocks_need > rlun->nr_free_blocks &&
- !list_empty(&rlun->prio_list)) {
- struct rrpc_block *rblk = block_prio_find_max(rlun);
-
- if (!rblk->nr_invalid_pages)
- break;
-
- gcb = mempool_alloc(rrpc->gcb_pool, GFP_ATOMIC);
- if (!gcb)
- break;
-
- list_del_init(&rblk->prio);
-
- WARN_ON(!block_is_full(rrpc, rblk));
-
- pr_debug("rrpc: selected block 'ch:%d,lun:%d,blk:%d' for GC\n",
- rlun->bppa.g.ch, rlun->bppa.g.lun,
- rblk->id);
-
- gcb->rrpc = rrpc;
- gcb->rblk = rblk;
- INIT_WORK(&gcb->ws_gc, rrpc_block_gc);
-
- queue_work(rrpc->kgc_wq, &gcb->ws_gc);
-
- nr_blocks_need--;
- }
- spin_unlock(&rlun->lock);
-
- /* TODO: Hint that request queue can be started again */
-}
-
-static void rrpc_gc_queue(struct work_struct *work)
-{
- struct rrpc_block_gc *gcb = container_of(work, struct rrpc_block_gc,
- ws_gc);
- struct rrpc *rrpc = gcb->rrpc;
- struct rrpc_block *rblk = gcb->rblk;
- struct rrpc_lun *rlun = rblk->rlun;
-
- spin_lock(&rlun->lock);
- list_add_tail(&rblk->prio, &rlun->prio_list);
- spin_unlock(&rlun->lock);
-
- mempool_free(gcb, rrpc->gcb_pool);
- pr_debug("nvm: block 'ch:%d,lun:%d,blk:%d' full, allow GC (sched)\n",
- rlun->bppa.g.ch, rlun->bppa.g.lun,
- rblk->id);
-}
-
-static const struct block_device_operations rrpc_fops = {
- .owner = THIS_MODULE,
-};
-
-static struct rrpc_lun *rrpc_get_lun_rr(struct rrpc *rrpc, int is_gc)
-{
- unsigned int i;
- struct rrpc_lun *rlun, *max_free;
-
- if (!is_gc)
- return get_next_lun(rrpc);
-
- /* during GC, we don't care about RR, instead we want to make
- * sure that we maintain evenness between the block luns.
- */
- max_free = &rrpc->luns[0];
- /* prevent GC-ing lun from devouring pages of a lun with
- * little free blocks. We don't take the lock as we only need an
- * estimate.
- */
- rrpc_for_each_lun(rrpc, rlun, i) {
- if (rlun->nr_free_blocks > max_free->nr_free_blocks)
- max_free = rlun;
- }
-
- return max_free;
-}
-
-static struct rrpc_addr *rrpc_update_map(struct rrpc *rrpc, sector_t laddr,
- struct rrpc_block *rblk, u64 paddr)
-{
- struct rrpc_addr *gp;
- struct rrpc_rev_addr *rev;
-
- BUG_ON(laddr >= rrpc->nr_sects);
-
- gp = &rrpc->trans_map[laddr];
- spin_lock(&rrpc->rev_lock);
- if (gp->rblk)
- rrpc_page_invalidate(rrpc, gp);
-
- gp->addr = paddr;
- gp->rblk = rblk;
-
- rev = &rrpc->rev_trans_map[gp->addr];
- rev->addr = laddr;
- spin_unlock(&rrpc->rev_lock);
-
- return gp;
-}
-
-static u64 rrpc_alloc_addr(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- u64 addr = ADDR_EMPTY;
-
- spin_lock(&rblk->lock);
- if (block_is_full(rrpc, rblk))
- goto out;
-
- addr = rblk->next_page;
-
- rblk->next_page++;
-out:
- spin_unlock(&rblk->lock);
- return addr;
-}
-
-/* Map logical address to a physical page. The mapping implements a round robin
- * approach and allocates a page from the next lun available.
- *
- * Returns rrpc_addr with the physical address and block. Returns NULL if no
- * blocks in the next rlun are available.
- */
-static struct ppa_addr rrpc_map_page(struct rrpc *rrpc, sector_t laddr,
- int is_gc)
-{
- struct nvm_tgt_dev *tgt_dev = rrpc->dev;
- struct rrpc_lun *rlun;
- struct rrpc_block *rblk, **cur_rblk;
- struct rrpc_addr *p;
- struct ppa_addr ppa;
- u64 paddr;
- int gc_force = 0;
-
- ppa.ppa = ADDR_EMPTY;
- rlun = rrpc_get_lun_rr(rrpc, is_gc);
-
- if (!is_gc && rlun->nr_free_blocks < rrpc->nr_luns * 4)
- return ppa;
-
- /*
- * page allocation steps:
- * 1. Try to allocate new page from current rblk
- * 2a. If succeed, proceed to map it in and return
- * 2b. If fail, first try to allocate a new block from media manger,
- * and then retry step 1. Retry until the normal block pool is
- * exhausted.
- * 3. If exhausted, and garbage collector is requesting the block,
- * go to the reserved block and retry step 1.
- * In the case that this fails as well, or it is not GC
- * requesting, report not able to retrieve a block and let the
- * caller handle further processing.
- */
-
- spin_lock(&rlun->lock);
- cur_rblk = &rlun->cur;
- rblk = rlun->cur;
-retry:
- paddr = rrpc_alloc_addr(rrpc, rblk);
-
- if (paddr != ADDR_EMPTY)
- goto done;
-
- if (!list_empty(&rlun->wblk_list)) {
-new_blk:
- rblk = list_first_entry(&rlun->wblk_list, struct rrpc_block,
- prio);
- rrpc_set_lun_cur(rlun, rblk, cur_rblk);
- list_del(&rblk->prio);
- goto retry;
- }
- spin_unlock(&rlun->lock);
-
- rblk = rrpc_get_blk(rrpc, rlun, gc_force);
- if (rblk) {
- spin_lock(&rlun->lock);
- list_add_tail(&rblk->prio, &rlun->wblk_list);
- /*
- * another thread might already have added a new block,
- * Therefore, make sure that one is used, instead of the
- * one just added.
- */
- goto new_blk;
- }
-
- if (unlikely(is_gc) && !gc_force) {
- /* retry from emergency gc block */
- cur_rblk = &rlun->gc_cur;
- rblk = rlun->gc_cur;
- gc_force = 1;
- spin_lock(&rlun->lock);
- goto retry;
- }
-
- pr_err("rrpc: failed to allocate new block\n");
- return ppa;
-done:
- spin_unlock(&rlun->lock);
- p = rrpc_update_map(rrpc, laddr, rblk, paddr);
- if (!p)
- return ppa;
-
- /* return global address */
- return rrpc_ppa_to_gaddr(tgt_dev, p);
-}
-
-static void rrpc_run_gc(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct rrpc_block_gc *gcb;
-
- gcb = mempool_alloc(rrpc->gcb_pool, GFP_ATOMIC);
- if (!gcb) {
- pr_err("rrpc: unable to queue block for gc.");
- return;
- }
-
- gcb->rrpc = rrpc;
- gcb->rblk = rblk;
-
- INIT_WORK(&gcb->ws_gc, rrpc_gc_queue);
- queue_work(rrpc->kgc_wq, &gcb->ws_gc);
-}
-
-static struct rrpc_lun *rrpc_ppa_to_lun(struct rrpc *rrpc, struct ppa_addr p)
-{
- struct rrpc_lun *rlun = NULL;
- int i;
-
- for (i = 0; i < rrpc->nr_luns; i++) {
- if (rrpc->luns[i].bppa.g.ch == p.g.ch &&
- rrpc->luns[i].bppa.g.lun == p.g.lun) {
- rlun = &rrpc->luns[i];
- break;
- }
- }
-
- return rlun;
-}
-
-static void __rrpc_mark_bad_block(struct rrpc *rrpc, struct ppa_addr ppa)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_lun *rlun;
- struct rrpc_block *rblk;
-
- rlun = rrpc_ppa_to_lun(rrpc, ppa);
- rblk = &rlun->blocks[ppa.g.blk];
- rblk->state = NVM_BLK_ST_BAD;
-
- nvm_set_tgt_bb_tbl(dev, &ppa, 1, NVM_BLK_T_GRWN_BAD);
-}
-
-static void rrpc_mark_bad_block(struct rrpc *rrpc, struct nvm_rq *rqd)
-{
- void *comp_bits = &rqd->ppa_status;
- struct ppa_addr ppa, prev_ppa;
- int nr_ppas = rqd->nr_ppas;
- int bit;
-
- if (rqd->nr_ppas == 1)
- __rrpc_mark_bad_block(rrpc, rqd->ppa_addr);
-
- ppa_set_empty(&prev_ppa);
- bit = -1;
- while ((bit = find_next_bit(comp_bits, nr_ppas, bit + 1)) < nr_ppas) {
- ppa = rqd->ppa_list[bit];
- if (ppa_cmp_blk(ppa, prev_ppa))
- continue;
-
- __rrpc_mark_bad_block(rrpc, ppa);
- }
-}
-
-static void rrpc_end_io_write(struct rrpc *rrpc, struct rrpc_rq *rrqd,
- sector_t laddr, uint8_t npages)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_addr *p;
- struct rrpc_block *rblk;
- int cmnt_size, i;
-
- for (i = 0; i < npages; i++) {
- p = &rrpc->trans_map[laddr + i];
- rblk = p->rblk;
-
- cmnt_size = atomic_inc_return(&rblk->data_cmnt_size);
- if (unlikely(cmnt_size == dev->geo.sec_per_blk))
- rrpc_run_gc(rrpc, rblk);
- }
-}
-
-static void rrpc_end_io(struct nvm_rq *rqd)
-{
- struct rrpc *rrpc = rqd->private;
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_rq *rrqd = nvm_rq_to_pdu(rqd);
- uint8_t npages = rqd->nr_ppas;
- sector_t laddr = rrpc_get_laddr(rqd->bio) - npages;
-
- if (bio_data_dir(rqd->bio) == WRITE) {
- if (rqd->error == NVM_RSP_ERR_FAILWRITE)
- rrpc_mark_bad_block(rrpc, rqd);
-
- rrpc_end_io_write(rrpc, rrqd, laddr, npages);
- }
-
- bio_put(rqd->bio);
-
- if (rrqd->flags & NVM_IOTYPE_GC)
- return;
-
- rrpc_unlock_rq(rrpc, rqd);
-
- if (npages > 1)
- nvm_dev_dma_free(dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
-
- mempool_free(rqd, rrpc->rq_pool);
-}
-
-static int rrpc_read_ppalist_rq(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags, int npages)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_inflight_rq *r = rrpc_get_inflight_rq(rqd);
- struct rrpc_addr *gp;
- sector_t laddr = rrpc_get_laddr(bio);
- int is_gc = flags & NVM_IOTYPE_GC;
- int i;
-
- if (!is_gc && rrpc_lock_rq(rrpc, bio, rqd)) {
- nvm_dev_dma_free(dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
- return NVM_IO_REQUEUE;
- }
-
- for (i = 0; i < npages; i++) {
- /* We assume that mapping occurs at 4KB granularity */
- BUG_ON(!(laddr + i < rrpc->nr_sects));
- gp = &rrpc->trans_map[laddr + i];
-
- if (gp->rblk) {
- rqd->ppa_list[i] = rrpc_ppa_to_gaddr(dev, gp);
- } else {
- BUG_ON(is_gc);
- rrpc_unlock_laddr(rrpc, r);
- nvm_dev_dma_free(dev->parent, rqd->ppa_list,
- rqd->dma_ppa_list);
- return NVM_IO_DONE;
- }
- }
-
- rqd->opcode = NVM_OP_HBREAD;
-
- return NVM_IO_OK;
-}
-
-static int rrpc_read_rq(struct rrpc *rrpc, struct bio *bio, struct nvm_rq *rqd,
- unsigned long flags)
-{
- int is_gc = flags & NVM_IOTYPE_GC;
- sector_t laddr = rrpc_get_laddr(bio);
- struct rrpc_addr *gp;
-
- if (!is_gc && rrpc_lock_rq(rrpc, bio, rqd))
- return NVM_IO_REQUEUE;
-
- BUG_ON(!(laddr < rrpc->nr_sects));
- gp = &rrpc->trans_map[laddr];
-
- if (gp->rblk) {
- rqd->ppa_addr = rrpc_ppa_to_gaddr(rrpc->dev, gp);
- } else {
- BUG_ON(is_gc);
- rrpc_unlock_rq(rrpc, rqd);
- return NVM_IO_DONE;
- }
-
- rqd->opcode = NVM_OP_HBREAD;
-
- return NVM_IO_OK;
-}
-
-static int rrpc_write_ppalist_rq(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags, int npages)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_inflight_rq *r = rrpc_get_inflight_rq(rqd);
- struct ppa_addr p;
- sector_t laddr = rrpc_get_laddr(bio);
- int is_gc = flags & NVM_IOTYPE_GC;
- int i;
-
- if (!is_gc && rrpc_lock_rq(rrpc, bio, rqd)) {
- nvm_dev_dma_free(dev->parent, rqd->ppa_list, rqd->dma_ppa_list);
- return NVM_IO_REQUEUE;
- }
-
- for (i = 0; i < npages; i++) {
- /* We assume that mapping occurs at 4KB granularity */
- p = rrpc_map_page(rrpc, laddr + i, is_gc);
- if (p.ppa == ADDR_EMPTY) {
- BUG_ON(is_gc);
- rrpc_unlock_laddr(rrpc, r);
- nvm_dev_dma_free(dev->parent, rqd->ppa_list,
- rqd->dma_ppa_list);
- rrpc_gc_kick(rrpc);
- return NVM_IO_REQUEUE;
- }
-
- rqd->ppa_list[i] = p;
- }
-
- rqd->opcode = NVM_OP_HBWRITE;
-
- return NVM_IO_OK;
-}
-
-static int rrpc_write_rq(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags)
-{
- struct ppa_addr p;
- int is_gc = flags & NVM_IOTYPE_GC;
- sector_t laddr = rrpc_get_laddr(bio);
-
- if (!is_gc && rrpc_lock_rq(rrpc, bio, rqd))
- return NVM_IO_REQUEUE;
-
- p = rrpc_map_page(rrpc, laddr, is_gc);
- if (p.ppa == ADDR_EMPTY) {
- BUG_ON(is_gc);
- rrpc_unlock_rq(rrpc, rqd);
- rrpc_gc_kick(rrpc);
- return NVM_IO_REQUEUE;
- }
-
- rqd->ppa_addr = p;
- rqd->opcode = NVM_OP_HBWRITE;
-
- return NVM_IO_OK;
-}
-
-static int rrpc_setup_rq(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags, uint8_t npages)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
-
- if (npages > 1) {
- rqd->ppa_list = nvm_dev_dma_alloc(dev->parent, GFP_KERNEL,
- &rqd->dma_ppa_list);
- if (!rqd->ppa_list) {
- pr_err("rrpc: not able to allocate ppa list\n");
- return NVM_IO_ERR;
- }
-
- if (bio_op(bio) == REQ_OP_WRITE)
- return rrpc_write_ppalist_rq(rrpc, bio, rqd, flags,
- npages);
-
- return rrpc_read_ppalist_rq(rrpc, bio, rqd, flags, npages);
- }
-
- if (bio_op(bio) == REQ_OP_WRITE)
- return rrpc_write_rq(rrpc, bio, rqd, flags);
-
- return rrpc_read_rq(rrpc, bio, rqd, flags);
-}
-
-static int rrpc_submit_io(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd, unsigned long flags)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_rq *rrq = nvm_rq_to_pdu(rqd);
- uint8_t nr_pages = rrpc_get_pages(bio);
- int bio_size = bio_sectors(bio) << 9;
- int err;
-
- if (bio_size < dev->geo.sec_size)
- return NVM_IO_ERR;
- else if (bio_size > dev->geo.max_rq_size)
- return NVM_IO_ERR;
-
- err = rrpc_setup_rq(rrpc, bio, rqd, flags, nr_pages);
- if (err)
- return err;
-
- bio_get(bio);
- rqd->bio = bio;
- rqd->private = rrpc;
- rqd->nr_ppas = nr_pages;
- rqd->end_io = rrpc_end_io;
- rrq->flags = flags;
-
- err = nvm_submit_io(dev, rqd);
- if (err) {
- pr_err("rrpc: I/O submission failed: %d\n", err);
- bio_put(bio);
- if (!(flags & NVM_IOTYPE_GC)) {
- rrpc_unlock_rq(rrpc, rqd);
- if (rqd->nr_ppas > 1)
- nvm_dev_dma_free(dev->parent, rqd->ppa_list,
- rqd->dma_ppa_list);
- }
- return NVM_IO_ERR;
- }
-
- return NVM_IO_OK;
-}
-
-static blk_qc_t rrpc_make_rq(struct request_queue *q, struct bio *bio)
-{
- struct rrpc *rrpc = q->queuedata;
- struct nvm_rq *rqd;
- int err;
-
- blk_queue_split(q, &bio);
-
- if (bio_op(bio) == REQ_OP_DISCARD) {
- rrpc_discard(rrpc, bio);
- return BLK_QC_T_NONE;
- }
-
- rqd = mempool_alloc(rrpc->rq_pool, GFP_KERNEL);
- memset(rqd, 0, sizeof(struct nvm_rq));
-
- err = rrpc_submit_io(rrpc, bio, rqd, NVM_IOTYPE_NONE);
- switch (err) {
- case NVM_IO_OK:
- return BLK_QC_T_NONE;
- case NVM_IO_ERR:
- bio_io_error(bio);
- break;
- case NVM_IO_DONE:
- bio_endio(bio);
- break;
- case NVM_IO_REQUEUE:
- spin_lock(&rrpc->bio_lock);
- bio_list_add(&rrpc->requeue_bios, bio);
- spin_unlock(&rrpc->bio_lock);
- queue_work(rrpc->kgc_wq, &rrpc->ws_requeue);
- break;
- }
-
- mempool_free(rqd, rrpc->rq_pool);
- return BLK_QC_T_NONE;
-}
-
-static void rrpc_requeue(struct work_struct *work)
-{
- struct rrpc *rrpc = container_of(work, struct rrpc, ws_requeue);
- struct bio_list bios;
- struct bio *bio;
-
- bio_list_init(&bios);
-
- spin_lock(&rrpc->bio_lock);
- bio_list_merge(&bios, &rrpc->requeue_bios);
- bio_list_init(&rrpc->requeue_bios);
- spin_unlock(&rrpc->bio_lock);
-
- while ((bio = bio_list_pop(&bios)))
- rrpc_make_rq(rrpc->disk->queue, bio);
-}
-
-static void rrpc_gc_free(struct rrpc *rrpc)
-{
- if (rrpc->krqd_wq)
- destroy_workqueue(rrpc->krqd_wq);
-
- if (rrpc->kgc_wq)
- destroy_workqueue(rrpc->kgc_wq);
-}
-
-static int rrpc_gc_init(struct rrpc *rrpc)
-{
- rrpc->krqd_wq = alloc_workqueue("rrpc-lun", WQ_MEM_RECLAIM|WQ_UNBOUND,
- rrpc->nr_luns);
- if (!rrpc->krqd_wq)
- return -ENOMEM;
-
- rrpc->kgc_wq = alloc_workqueue("rrpc-bg", WQ_MEM_RECLAIM, 1);
- if (!rrpc->kgc_wq)
- return -ENOMEM;
-
- timer_setup(&rrpc->gc_timer, rrpc_gc_timer, 0);
-
- return 0;
-}
-
-static void rrpc_map_free(struct rrpc *rrpc)
-{
- vfree(rrpc->rev_trans_map);
- vfree(rrpc->trans_map);
-}
-
-static int rrpc_l2p_update(u64 slba, u32 nlb, __le64 *entries, void *private)
-{
- struct rrpc *rrpc = (struct rrpc *)private;
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_addr *addr = rrpc->trans_map + slba;
- struct rrpc_rev_addr *raddr = rrpc->rev_trans_map;
- struct rrpc_lun *rlun;
- struct rrpc_block *rblk;
- u64 i;
-
- for (i = 0; i < nlb; i++) {
- struct ppa_addr gaddr;
- u64 pba = le64_to_cpu(entries[i]);
- unsigned int mod;
-
- /* LNVM treats address-spaces as silos, LBA and PBA are
- * equally large and zero-indexed.
- */
- if (unlikely(pba >= dev->total_secs && pba != U64_MAX)) {
- pr_err("nvm: L2P data entry is out of bounds!\n");
- pr_err("nvm: Maybe loaded an old target L2P\n");
- return -EINVAL;
- }
-
- /* Address zero is a special one. The first page on a disk is
- * protected. As it often holds internal device boot
- * information.
- */
- if (!pba)
- continue;
-
- div_u64_rem(pba, rrpc->nr_sects, &mod);
-
- gaddr = rrpc_recov_addr(dev, pba);
- rlun = rrpc_ppa_to_lun(rrpc, gaddr);
- if (!rlun) {
- pr_err("rrpc: l2p corruption on lba %llu\n",
- slba + i);
- return -EINVAL;
- }
-
- rblk = &rlun->blocks[gaddr.g.blk];
- if (!rblk->state) {
- /* 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. The block is assumed to be open.
- */
- list_move_tail(&rblk->list, &rlun->used_list);
- rblk->state = NVM_BLK_ST_TGT;
- rlun->nr_free_blocks--;
- }
-
- addr[i].addr = pba;
- addr[i].rblk = rblk;
- raddr[mod].addr = slba + i;
- }
-
- return 0;
-}
-
-static int rrpc_map_init(struct rrpc *rrpc)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- sector_t i;
- int ret;
-
- rrpc->trans_map = vzalloc(sizeof(struct rrpc_addr) * rrpc->nr_sects);
- if (!rrpc->trans_map)
- return -ENOMEM;
-
- rrpc->rev_trans_map = vmalloc(sizeof(struct rrpc_rev_addr)
- * rrpc->nr_sects);
- if (!rrpc->rev_trans_map)
- return -ENOMEM;
-
- for (i = 0; i < rrpc->nr_sects; i++) {
- struct rrpc_addr *p = &rrpc->trans_map[i];
- struct rrpc_rev_addr *r = &rrpc->rev_trans_map[i];
-
- p->addr = ADDR_EMPTY;
- r->addr = ADDR_EMPTY;
- }
-
- /* Bring up the mapping table from device */
- ret = nvm_get_l2p_tbl(dev, rrpc->soffset, rrpc->nr_sects,
- rrpc_l2p_update, rrpc);
- if (ret) {
- pr_err("nvm: rrpc: could not read L2P table.\n");
- return -EINVAL;
- }
-
- return 0;
-}
-
-/* Minimum pages needed within a lun */
-#define PAGE_POOL_SIZE 16
-#define ADDR_POOL_SIZE 64
-
-static int rrpc_core_init(struct rrpc *rrpc)
-{
- down_write(&rrpc_lock);
- if (!rrpc_gcb_cache) {
- rrpc_gcb_cache = kmem_cache_create("rrpc_gcb",
- sizeof(struct rrpc_block_gc), 0, 0, NULL);
- if (!rrpc_gcb_cache) {
- up_write(&rrpc_lock);
- return -ENOMEM;
- }
-
- rrpc_rq_cache = kmem_cache_create("rrpc_rq",
- sizeof(struct nvm_rq) + sizeof(struct rrpc_rq),
- 0, 0, NULL);
- if (!rrpc_rq_cache) {
- kmem_cache_destroy(rrpc_gcb_cache);
- up_write(&rrpc_lock);
- return -ENOMEM;
- }
- }
- up_write(&rrpc_lock);
-
- rrpc->page_pool = mempool_create_page_pool(PAGE_POOL_SIZE, 0);
- if (!rrpc->page_pool)
- return -ENOMEM;
-
- rrpc->gcb_pool = mempool_create_slab_pool(rrpc->dev->geo.nr_luns,
- rrpc_gcb_cache);
- if (!rrpc->gcb_pool)
- return -ENOMEM;
-
- rrpc->rq_pool = mempool_create_slab_pool(64, rrpc_rq_cache);
- if (!rrpc->rq_pool)
- return -ENOMEM;
-
- spin_lock_init(&rrpc->inflights.lock);
- INIT_LIST_HEAD(&rrpc->inflights.reqs);
-
- return 0;
-}
-
-static void rrpc_core_free(struct rrpc *rrpc)
-{
- mempool_destroy(rrpc->page_pool);
- mempool_destroy(rrpc->gcb_pool);
- mempool_destroy(rrpc->rq_pool);
-}
-
-static void rrpc_luns_free(struct rrpc *rrpc)
-{
- struct rrpc_lun *rlun;
- int i;
-
- if (!rrpc->luns)
- return;
-
- for (i = 0; i < rrpc->nr_luns; i++) {
- rlun = &rrpc->luns[i];
- vfree(rlun->blocks);
- }
-
- kfree(rrpc->luns);
-}
-
-static int rrpc_bb_discovery(struct nvm_tgt_dev *dev, struct rrpc_lun *rlun)
-{
- struct nvm_geo *geo = &dev->geo;
- struct rrpc_block *rblk;
- struct ppa_addr ppa;
- u8 *blks;
- int nr_blks;
- int i;
- int ret;
-
- if (!dev->parent->ops->get_bb_tbl)
- return 0;
-
- nr_blks = geo->blks_per_lun * geo->plane_mode;
- blks = kmalloc(nr_blks, GFP_KERNEL);
- if (!blks)
- return -ENOMEM;
-
- ppa.ppa = 0;
- ppa.g.ch = rlun->bppa.g.ch;
- ppa.g.lun = rlun->bppa.g.lun;
-
- ret = nvm_get_tgt_bb_tbl(dev, ppa, blks);
- if (ret) {
- pr_err("rrpc: could not get BB table\n");
- goto out;
- }
-
- nr_blks = nvm_bb_tbl_fold(dev->parent, blks, nr_blks);
- if (nr_blks < 0) {
- ret = nr_blks;
- goto out;
- }
-
- for (i = 0; i < nr_blks; i++) {
- if (blks[i] == NVM_BLK_T_FREE)
- continue;
-
- rblk = &rlun->blocks[i];
- list_move_tail(&rblk->list, &rlun->bb_list);
- rblk->state = NVM_BLK_ST_BAD;
- rlun->nr_free_blocks--;
- }
-
-out:
- kfree(blks);
- return ret;
-}
-
-static void rrpc_set_lun_ppa(struct rrpc_lun *rlun, struct ppa_addr ppa)
-{
- rlun->bppa.ppa = 0;
- rlun->bppa.g.ch = ppa.g.ch;
- rlun->bppa.g.lun = ppa.g.lun;
-}
-
-static int rrpc_luns_init(struct rrpc *rrpc, struct ppa_addr *luns)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct nvm_geo *geo = &dev->geo;
- struct rrpc_lun *rlun;
- int i, j, ret = -EINVAL;
-
- if (geo->sec_per_blk > MAX_INVALID_PAGES_STORAGE * BITS_PER_LONG) {
- pr_err("rrpc: number of pages per block too high.");
- return -EINVAL;
- }
-
- spin_lock_init(&rrpc->rev_lock);
-
- rrpc->luns = kcalloc(rrpc->nr_luns, sizeof(struct rrpc_lun),
- GFP_KERNEL);
- if (!rrpc->luns)
- return -ENOMEM;
-
- /* 1:1 mapping */
- for (i = 0; i < rrpc->nr_luns; i++) {
- rlun = &rrpc->luns[i];
- rlun->id = i;
- rrpc_set_lun_ppa(rlun, luns[i]);
- rlun->blocks = vzalloc(sizeof(struct rrpc_block) *
- geo->blks_per_lun);
- if (!rlun->blocks) {
- ret = -ENOMEM;
- goto err;
- }
-
- INIT_LIST_HEAD(&rlun->free_list);
- INIT_LIST_HEAD(&rlun->used_list);
- INIT_LIST_HEAD(&rlun->bb_list);
-
- for (j = 0; j < geo->blks_per_lun; j++) {
- struct rrpc_block *rblk = &rlun->blocks[j];
-
- rblk->id = j;
- rblk->rlun = rlun;
- rblk->state = NVM_BLK_T_FREE;
- INIT_LIST_HEAD(&rblk->prio);
- INIT_LIST_HEAD(&rblk->list);
- spin_lock_init(&rblk->lock);
-
- list_add_tail(&rblk->list, &rlun->free_list);
- }
-
- rlun->rrpc = rrpc;
- rlun->nr_free_blocks = geo->blks_per_lun;
- rlun->reserved_blocks = 2; /* for GC only */
-
- INIT_LIST_HEAD(&rlun->prio_list);
- INIT_LIST_HEAD(&rlun->wblk_list);
-
- INIT_WORK(&rlun->ws_gc, rrpc_lun_gc);
- spin_lock_init(&rlun->lock);
-
- if (rrpc_bb_discovery(dev, rlun))
- goto err;
-
- }
-
- return 0;
-err:
- return ret;
-}
-
-/* returns 0 on success and stores the beginning address in *begin */
-static int rrpc_area_init(struct rrpc *rrpc, sector_t *begin)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- sector_t size = rrpc->nr_sects * dev->geo.sec_size;
- int ret;
-
- size >>= 9;
-
- ret = nvm_get_area(dev, begin, size);
- if (!ret)
- *begin >>= (ilog2(dev->geo.sec_size) - 9);
-
- return ret;
-}
-
-static void rrpc_area_free(struct rrpc *rrpc)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- sector_t begin = rrpc->soffset << (ilog2(dev->geo.sec_size) - 9);
-
- nvm_put_area(dev, begin);
-}
-
-static void rrpc_free(struct rrpc *rrpc)
-{
- rrpc_gc_free(rrpc);
- rrpc_map_free(rrpc);
- rrpc_core_free(rrpc);
- rrpc_luns_free(rrpc);
- rrpc_area_free(rrpc);
-
- kfree(rrpc);
-}
-
-static void rrpc_exit(void *private)
-{
- struct rrpc *rrpc = private;
-
- del_timer(&rrpc->gc_timer);
-
- flush_workqueue(rrpc->krqd_wq);
- flush_workqueue(rrpc->kgc_wq);
-
- rrpc_free(rrpc);
-}
-
-static sector_t rrpc_capacity(void *private)
-{
- struct rrpc *rrpc = private;
- struct nvm_tgt_dev *dev = rrpc->dev;
- sector_t reserved, provisioned;
-
- /* cur, gc, and two emergency blocks for each lun */
- reserved = rrpc->nr_luns * dev->geo.sec_per_blk * 4;
- provisioned = rrpc->nr_sects - reserved;
-
- if (reserved > rrpc->nr_sects) {
- pr_err("rrpc: not enough space available to expose storage.\n");
- return 0;
- }
-
- sector_div(provisioned, 10);
- return provisioned * 9 * NR_PHY_IN_LOG;
-}
-
-/*
- * Looks up the logical address from reverse trans map and check if its valid by
- * comparing the logical to physical address with the physical address.
- * Returns 0 on free, otherwise 1 if in use
- */
-static void rrpc_block_map_update(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- int offset;
- struct rrpc_addr *laddr;
- u64 bpaddr, paddr, pladdr;
-
- bpaddr = block_to_rel_addr(rrpc, rblk);
- for (offset = 0; offset < dev->geo.sec_per_blk; offset++) {
- paddr = bpaddr + offset;
-
- pladdr = rrpc->rev_trans_map[paddr].addr;
- if (pladdr == ADDR_EMPTY)
- continue;
-
- laddr = &rrpc->trans_map[pladdr];
-
- if (paddr == laddr->addr) {
- laddr->rblk = rblk;
- } else {
- set_bit(offset, rblk->invalid_pages);
- rblk->nr_invalid_pages++;
- }
- }
-}
-
-static int rrpc_blocks_init(struct rrpc *rrpc)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct rrpc_lun *rlun;
- struct rrpc_block *rblk;
- int lun_iter, blk_iter;
-
- for (lun_iter = 0; lun_iter < rrpc->nr_luns; lun_iter++) {
- rlun = &rrpc->luns[lun_iter];
-
- for (blk_iter = 0; blk_iter < dev->geo.blks_per_lun;
- blk_iter++) {
- rblk = &rlun->blocks[blk_iter];
- rrpc_block_map_update(rrpc, rblk);
- }
- }
-
- return 0;
-}
-
-static int rrpc_luns_configure(struct rrpc *rrpc)
-{
- struct rrpc_lun *rlun;
- struct rrpc_block *rblk;
- int i;
-
- for (i = 0; i < rrpc->nr_luns; i++) {
- rlun = &rrpc->luns[i];
-
- rblk = rrpc_get_blk(rrpc, rlun, 0);
- if (!rblk)
- goto err;
- rrpc_set_lun_cur(rlun, rblk, &rlun->cur);
-
- /* Emergency gc block */
- rblk = rrpc_get_blk(rrpc, rlun, 1);
- if (!rblk)
- goto err;
- rrpc_set_lun_cur(rlun, rblk, &rlun->gc_cur);
- }
-
- return 0;
-err:
- rrpc_put_blks(rrpc);
- return -EINVAL;
-}
-
-static struct nvm_tgt_type tt_rrpc;
-
-static void *rrpc_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
- int flags)
-{
- struct request_queue *bqueue = dev->q;
- struct request_queue *tqueue = tdisk->queue;
- struct nvm_geo *geo = &dev->geo;
- struct rrpc *rrpc;
- sector_t soffset;
- int ret;
-
- if (!(dev->identity.dom & NVM_RSP_L2P)) {
- pr_err("nvm: rrpc: device does not support l2p (%x)\n",
- dev->identity.dom);
- return ERR_PTR(-EINVAL);
- }
-
- rrpc = kzalloc(sizeof(struct rrpc), GFP_KERNEL);
- if (!rrpc)
- return ERR_PTR(-ENOMEM);
-
- rrpc->dev = dev;
- rrpc->disk = tdisk;
-
- bio_list_init(&rrpc->requeue_bios);
- spin_lock_init(&rrpc->bio_lock);
- INIT_WORK(&rrpc->ws_requeue, rrpc_requeue);
-
- rrpc->nr_luns = geo->nr_luns;
- rrpc->nr_sects = (unsigned long long)geo->sec_per_lun * rrpc->nr_luns;
-
- /* simple round-robin strategy */
- atomic_set(&rrpc->next_lun, -1);
-
- ret = rrpc_area_init(rrpc, &soffset);
- if (ret < 0) {
- pr_err("nvm: rrpc: could not initialize area\n");
- return ERR_PTR(ret);
- }
- rrpc->soffset = soffset;
-
- ret = rrpc_luns_init(rrpc, dev->luns);
- if (ret) {
- pr_err("nvm: rrpc: could not initialize luns\n");
- goto err;
- }
-
- ret = rrpc_core_init(rrpc);
- if (ret) {
- pr_err("nvm: rrpc: could not initialize core\n");
- goto err;
- }
-
- ret = rrpc_map_init(rrpc);
- if (ret) {
- pr_err("nvm: rrpc: could not initialize maps\n");
- goto err;
- }
-
- ret = rrpc_blocks_init(rrpc);
- if (ret) {
- pr_err("nvm: rrpc: could not initialize state for blocks\n");
- goto err;
- }
-
- ret = rrpc_luns_configure(rrpc);
- if (ret) {
- pr_err("nvm: rrpc: not enough blocks available in LUNs.\n");
- goto err;
- }
-
- ret = rrpc_gc_init(rrpc);
- if (ret) {
- pr_err("nvm: rrpc: could not initialize gc\n");
- goto err;
- }
-
- /* inherit the size from the underlying device */
- blk_queue_logical_block_size(tqueue, queue_physical_block_size(bqueue));
- blk_queue_max_hw_sectors(tqueue, queue_max_hw_sectors(bqueue));
-
- pr_info("nvm: rrpc initialized with %u luns and %llu pages.\n",
- rrpc->nr_luns, (unsigned long long)rrpc->nr_sects);
-
- mod_timer(&rrpc->gc_timer, jiffies + msecs_to_jiffies(10));
-
- return rrpc;
-err:
- rrpc_free(rrpc);
- return ERR_PTR(ret);
-}
-
-/* round robin, page-based FTL, and cost-based GC */
-static struct nvm_tgt_type tt_rrpc = {
- .name = "rrpc",
- .version = {1, 0, 0},
-
- .make_rq = rrpc_make_rq,
- .capacity = rrpc_capacity,
-
- .init = rrpc_init,
- .exit = rrpc_exit,
-};
-
-static int __init rrpc_module_init(void)
-{
- return nvm_register_tgt_type(&tt_rrpc);
-}
-
-static void rrpc_module_exit(void)
-{
- nvm_unregister_tgt_type(&tt_rrpc);
-}
-
-module_init(rrpc_module_init);
-module_exit(rrpc_module_exit);
-MODULE_LICENSE("GPL v2");
-MODULE_DESCRIPTION("Block-Device Target for Open-Channel SSDs");
diff --git a/drivers/lightnvm/rrpc.h b/drivers/lightnvm/rrpc.h
deleted file mode 100644
index fdb6ff9..0000000
--- a/drivers/lightnvm/rrpc.h
+++ /dev/null
@@ -1,290 +0,0 @@
-/*
- * Copyright (C) 2015 IT University of Copenhagen
- * Initial release: Matias Bjorling <m@bjorling.me>
- *
- * 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.
- *
- * Implementation of a Round-robin page-based Hybrid FTL for Open-channel SSDs.
- */
-
-#ifndef RRPC_H_
-#define RRPC_H_
-
-#include <linux/blkdev.h>
-#include <linux/blk-mq.h>
-#include <linux/bio.h>
-#include <linux/module.h>
-#include <linux/kthread.h>
-#include <linux/vmalloc.h>
-
-#include <linux/lightnvm.h>
-
-/* Run only GC if less than 1/X blocks are free */
-#define GC_LIMIT_INVERSE 10
-#define GC_TIME_SECS 100
-
-#define RRPC_SECTOR (512)
-#define RRPC_EXPOSED_PAGE_SIZE (4096)
-
-#define NR_PHY_IN_LOG (RRPC_EXPOSED_PAGE_SIZE / RRPC_SECTOR)
-
-struct rrpc_inflight {
- struct list_head reqs;
- spinlock_t lock;
-};
-
-struct rrpc_inflight_rq {
- struct list_head list;
- sector_t l_start;
- sector_t l_end;
-};
-
-struct rrpc_rq {
- struct rrpc_inflight_rq inflight_rq;
- unsigned long flags;
-};
-
-struct rrpc_block {
- int id; /* id inside of LUN */
- struct rrpc_lun *rlun;
-
- struct list_head prio; /* LUN CG list */
- struct list_head list; /* LUN free, used, bb list */
-
-#define MAX_INVALID_PAGES_STORAGE 8
- /* Bitmap for invalid page intries */
- unsigned long invalid_pages[MAX_INVALID_PAGES_STORAGE];
- /* points to the next writable page within a block */
- unsigned int next_page;
- /* number of pages that are invalid, wrt host page size */
- unsigned int nr_invalid_pages;
-
- int state;
-
- spinlock_t lock;
- atomic_t data_cmnt_size; /* data pages committed to stable storage */
-};
-
-struct rrpc_lun {
- struct rrpc *rrpc;
-
- int id;
- struct ppa_addr bppa;
-
- struct rrpc_block *cur, *gc_cur;
- struct rrpc_block *blocks; /* Reference to block allocation */
-
- struct list_head prio_list; /* Blocks that may be GC'ed */
- struct list_head wblk_list; /* Queued blocks to be written to */
-
- /* 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
- */
- unsigned int nr_free_blocks; /* Number of unused blocks */
-
- struct work_struct ws_gc;
-
- int reserved_blocks;
-
- spinlock_t lock;
-};
-
-struct rrpc {
- struct nvm_tgt_dev *dev;
- struct gendisk *disk;
-
- sector_t soffset; /* logical sector offset */
-
- int nr_luns;
- struct rrpc_lun *luns;
-
- /* calculated values */
- unsigned long long nr_sects;
-
- /* Write strategy variables. Move these into each for structure for each
- * strategy
- */
- atomic_t next_lun; /* Whenever a page is written, this is updated
- * to point to the next write lun
- */
-
- spinlock_t bio_lock;
- struct bio_list requeue_bios;
- struct work_struct ws_requeue;
-
- /* Simple translation map of logical addresses to physical addresses.
- * The logical addresses is known by the host system, while the physical
- * addresses are used when writing to the disk block device.
- */
- struct rrpc_addr *trans_map;
- /* also store a reverse map for garbage collection */
- struct rrpc_rev_addr *rev_trans_map;
- spinlock_t rev_lock;
-
- struct rrpc_inflight inflights;
-
- mempool_t *addr_pool;
- mempool_t *page_pool;
- mempool_t *gcb_pool;
- mempool_t *rq_pool;
-
- struct timer_list gc_timer;
- struct workqueue_struct *krqd_wq;
- struct workqueue_struct *kgc_wq;
-};
-
-struct rrpc_block_gc {
- struct rrpc *rrpc;
- struct rrpc_block *rblk;
- struct work_struct ws_gc;
-};
-
-/* Logical to physical mapping */
-struct rrpc_addr {
- u64 addr;
- struct rrpc_block *rblk;
-};
-
-/* Physical to logical mapping */
-struct rrpc_rev_addr {
- u64 addr;
-};
-
-static inline struct ppa_addr rrpc_linear_to_generic_addr(struct nvm_geo *geo,
- struct ppa_addr r)
-{
- struct ppa_addr l;
- int secs, pgs;
- sector_t ppa = r.ppa;
-
- l.ppa = 0;
-
- div_u64_rem(ppa, geo->sec_per_pg, &secs);
- l.g.sec = secs;
-
- sector_div(ppa, geo->sec_per_pg);
- div_u64_rem(ppa, geo->pgs_per_blk, &pgs);
- l.g.pg = pgs;
-
- return l;
-}
-
-static inline struct ppa_addr rrpc_recov_addr(struct nvm_tgt_dev *dev, u64 pba)
-{
- return linear_to_generic_addr(&dev->geo, pba);
-}
-
-static inline u64 rrpc_blk_to_ppa(struct rrpc *rrpc, struct rrpc_block *rblk)
-{
- struct nvm_tgt_dev *dev = rrpc->dev;
- struct nvm_geo *geo = &dev->geo;
- struct rrpc_lun *rlun = rblk->rlun;
-
- return (rlun->id * geo->sec_per_lun) + (rblk->id * geo->sec_per_blk);
-}
-
-static inline sector_t rrpc_get_laddr(struct bio *bio)
-{
- return bio->bi_iter.bi_sector / NR_PHY_IN_LOG;
-}
-
-static inline unsigned int rrpc_get_pages(struct bio *bio)
-{
- return bio->bi_iter.bi_size / RRPC_EXPOSED_PAGE_SIZE;
-}
-
-static inline sector_t rrpc_get_sector(sector_t laddr)
-{
- return laddr * NR_PHY_IN_LOG;
-}
-
-static inline int request_intersects(struct rrpc_inflight_rq *r,
- sector_t laddr_start, sector_t laddr_end)
-{
- return (laddr_end >= r->l_start) && (laddr_start <= r->l_end);
-}
-
-static int __rrpc_lock_laddr(struct rrpc *rrpc, sector_t laddr,
- unsigned int pages, struct rrpc_inflight_rq *r)
-{
- sector_t laddr_end = laddr + pages - 1;
- struct rrpc_inflight_rq *rtmp;
-
- WARN_ON(irqs_disabled());
-
- spin_lock_irq(&rrpc->inflights.lock);
- list_for_each_entry(rtmp, &rrpc->inflights.reqs, list) {
- if (unlikely(request_intersects(rtmp, laddr, laddr_end))) {
- /* existing, overlapping request, come back later */
- spin_unlock_irq(&rrpc->inflights.lock);
- return 1;
- }
- }
-
- r->l_start = laddr;
- r->l_end = laddr_end;
-
- list_add_tail(&r->list, &rrpc->inflights.reqs);
- spin_unlock_irq(&rrpc->inflights.lock);
- return 0;
-}
-
-static inline int rrpc_lock_laddr(struct rrpc *rrpc, sector_t laddr,
- unsigned int pages,
- struct rrpc_inflight_rq *r)
-{
- BUG_ON((laddr + pages) > rrpc->nr_sects);
-
- return __rrpc_lock_laddr(rrpc, laddr, pages, r);
-}
-
-static inline struct rrpc_inflight_rq *rrpc_get_inflight_rq(struct nvm_rq *rqd)
-{
- struct rrpc_rq *rrqd = nvm_rq_to_pdu(rqd);
-
- return &rrqd->inflight_rq;
-}
-
-static inline int rrpc_lock_rq(struct rrpc *rrpc, struct bio *bio,
- struct nvm_rq *rqd)
-{
- sector_t laddr = rrpc_get_laddr(bio);
- unsigned int pages = rrpc_get_pages(bio);
- struct rrpc_inflight_rq *r = rrpc_get_inflight_rq(rqd);
-
- return rrpc_lock_laddr(rrpc, laddr, pages, r);
-}
-
-static inline void rrpc_unlock_laddr(struct rrpc *rrpc,
- struct rrpc_inflight_rq *r)
-{
- unsigned long flags;
-
- spin_lock_irqsave(&rrpc->inflights.lock, flags);
- list_del_init(&r->list);
- spin_unlock_irqrestore(&rrpc->inflights.lock, flags);
-}
-
-static inline void rrpc_unlock_rq(struct rrpc *rrpc, struct nvm_rq *rqd)
-{
- struct rrpc_inflight_rq *r = rrpc_get_inflight_rq(rqd);
- uint8_t pages = rqd->nr_ppas;
-
- BUG_ON((r->l_start + pages) > rrpc->nr_sects);
-
- rrpc_unlock_laddr(rrpc, r);
-}
-
-#endif /* RRPC_H_ */
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 03/25] lightnvm: use internal pblk methods
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
2018-01-05 13:15 ` [GIT PULL 01/25] null_blk: remove lightnvm support Matias Bjørling
2018-01-05 13:15 ` [GIT PULL 02/25] lightnvm: remove rrpc Matias Bjørling
@ 2018-01-05 13:15 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 04/25] lightnvm: remove hybrid ocssd 1.2 support Matias Bjørling
` (22 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:15 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
Now that rrpc has been removed, the only users of the ppa helpers
is pblk. However, pblk already defines similar functions.
Switch pblk to use the internal ones, and remove the generic ppa
helpers.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-map.c | 2 +-
drivers/lightnvm/pblk-write.c | 4 ++--
include/linux/lightnvm.h | 19 -------------------
3 files changed, 3 insertions(+), 22 deletions(-)
diff --git a/drivers/lightnvm/pblk-map.c b/drivers/lightnvm/pblk-map.c
index 6f3ecde..7445e64 100644
--- a/drivers/lightnvm/pblk-map.c
+++ b/drivers/lightnvm/pblk-map.c
@@ -146,7 +146,7 @@ void pblk_map_erase_rq(struct pblk *pblk, struct nvm_rq *rqd,
return;
/* Erase blocks that are bad in this line but might not be in next */
- if (unlikely(ppa_empty(*erase_ppa)) &&
+ if (unlikely(pblk_ppa_empty(*erase_ppa)) &&
bitmap_weight(d_line->blk_bitmap, lm->blk_per_line)) {
int bit = -1;
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 6c1cafa..6c30b7a 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -439,7 +439,7 @@ static int pblk_submit_io_set(struct pblk *pblk, struct nvm_rq *rqd)
struct pblk_line *meta_line;
int err;
- ppa_set_empty(&erase_ppa);
+ pblk_ppa_set_empty(&erase_ppa);
/* Assign lbas to ppas and populate request structure */
err = pblk_setup_w_rq(pblk, rqd, &erase_ppa);
@@ -457,7 +457,7 @@ static int pblk_submit_io_set(struct pblk *pblk, struct nvm_rq *rqd)
return NVM_IO_ERR;
}
- if (!ppa_empty(erase_ppa)) {
+ if (!pblk_ppa_empty(erase_ppa)) {
/* Submit erase for next data line */
if (pblk_blk_erase_async(pblk, erase_ppa)) {
struct pblk_line *e_line = pblk_line_get_erase(pblk);
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 2d1d9de..14e274b 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -418,25 +418,6 @@ static inline struct ppa_addr dev_to_generic_addr(struct nvm_tgt_dev *tgt_dev,
return l;
}
-static inline int ppa_empty(struct ppa_addr ppa_addr)
-{
- return (ppa_addr.ppa == ADDR_EMPTY);
-}
-
-static inline void ppa_set_empty(struct ppa_addr *ppa_addr)
-{
- ppa_addr->ppa = ADDR_EMPTY;
-}
-
-static inline int ppa_cmp_blk(struct ppa_addr ppa1, struct ppa_addr ppa2)
-{
- if (ppa_empty(ppa1) || ppa_empty(ppa2))
- return 0;
-
- return ((ppa1.g.ch == ppa2.g.ch) && (ppa1.g.lun == ppa2.g.lun) &&
- (ppa1.g.blk == ppa2.g.blk));
-}
-
typedef blk_qc_t (nvm_tgt_make_rq_fn)(struct request_queue *, struct bio *);
typedef sector_t (nvm_tgt_capacity_fn)(void *);
typedef void *(nvm_tgt_init_fn)(struct nvm_tgt_dev *, struct gendisk *,
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 04/25] lightnvm: remove hybrid ocssd 1.2 support
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (2 preceding siblings ...)
2018-01-05 13:15 ` [GIT PULL 03/25] lightnvm: use internal pblk methods Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 05/25] lightnvm: remove unnecessary field from nvm_rq Matias Bjørling
` (21 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
Now that rrpc have been removed. Also remove the hybrid 1.2 support
from the core.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 141 -------------------------------------------
drivers/nvme/host/lightnvm.c | 96 -----------------------------
include/linux/lightnvm.h | 43 -------------
3 files changed, 280 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 83249b4..390d5ef 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -45,12 +45,6 @@ struct nvm_dev_map {
int nr_chnls;
};
-struct nvm_area {
- struct list_head list;
- sector_t begin;
- sector_t end; /* end is excluded */
-};
-
static struct nvm_target *nvm_find_target(struct nvm_dev *dev, const char *name)
{
struct nvm_target *tgt;
@@ -524,35 +518,6 @@ static void nvm_rq_dev_to_tgt(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
nvm_ppa_dev_to_tgt(tgt_dev, rqd->ppa_list, rqd->nr_ppas);
}
-void nvm_part_to_tgt(struct nvm_dev *dev, sector_t *entries,
- int len)
-{
- struct nvm_geo *geo = &dev->geo;
- struct nvm_dev_map *dev_rmap = dev->rmap;
- u64 i;
-
- for (i = 0; i < len; i++) {
- struct nvm_ch_map *ch_rmap;
- int *lun_roffs;
- struct ppa_addr gaddr;
- u64 pba = le64_to_cpu(entries[i]);
- u64 diff;
-
- if (!pba)
- continue;
-
- gaddr = linear_to_generic_addr(geo, pba);
- ch_rmap = &dev_rmap->chnls[gaddr.g.ch];
- lun_roffs = ch_rmap->lun_offs;
-
- diff = ((ch_rmap->ch_off * geo->luns_per_chnl) +
- (lun_roffs[gaddr.g.lun])) * geo->sec_per_lun;
-
- entries[i] -= cpu_to_le64(diff);
- }
-}
-EXPORT_SYMBOL(nvm_part_to_tgt);
-
int nvm_register_tgt_type(struct nvm_tgt_type *tt)
{
int ret = 0;
@@ -726,112 +691,6 @@ int nvm_submit_io_sync(struct nvm_tgt_dev *tgt_dev, struct nvm_rq *rqd)
}
EXPORT_SYMBOL(nvm_submit_io_sync);
-int nvm_erase_sync(struct nvm_tgt_dev *tgt_dev, struct ppa_addr *ppas,
- int nr_ppas)
-{
- struct nvm_geo *geo = &tgt_dev->geo;
- struct nvm_rq rqd;
- int ret;
-
- memset(&rqd, 0, sizeof(struct nvm_rq));
-
- rqd.opcode = NVM_OP_ERASE;
- rqd.flags = geo->plane_mode >> 1;
-
- ret = nvm_set_rqd_ppalist(tgt_dev, &rqd, ppas, nr_ppas);
- if (ret)
- return ret;
-
- ret = nvm_submit_io_sync(tgt_dev, &rqd);
- if (ret) {
- pr_err("rrpr: erase I/O submission failed: %d\n", ret);
- goto free_ppa_list;
- }
-
-free_ppa_list:
- nvm_free_rqd_ppalist(tgt_dev, &rqd);
-
- return ret;
-}
-EXPORT_SYMBOL(nvm_erase_sync);
-
-int nvm_get_l2p_tbl(struct nvm_tgt_dev *tgt_dev, u64 slba, u32 nlb,
- nvm_l2p_update_fn *update_l2p, void *priv)
-{
- struct nvm_dev *dev = tgt_dev->parent;
-
- if (!dev->ops->get_l2p_tbl)
- return 0;
-
- return dev->ops->get_l2p_tbl(dev, slba, nlb, update_l2p, priv);
-}
-EXPORT_SYMBOL(nvm_get_l2p_tbl);
-
-int nvm_get_area(struct nvm_tgt_dev *tgt_dev, sector_t *lba, sector_t len)
-{
- struct nvm_dev *dev = tgt_dev->parent;
- struct nvm_geo *geo = &dev->geo;
- struct nvm_area *area, *prev, *next;
- sector_t begin = 0;
- sector_t max_sectors = (geo->sec_size * dev->total_secs) >> 9;
-
- if (len > max_sectors)
- return -EINVAL;
-
- area = kmalloc(sizeof(struct nvm_area), GFP_KERNEL);
- if (!area)
- return -ENOMEM;
-
- prev = NULL;
-
- spin_lock(&dev->lock);
- list_for_each_entry(next, &dev->area_list, list) {
- if (begin + len > next->begin) {
- begin = next->end;
- prev = next;
- continue;
- }
- break;
- }
-
- if ((begin + len) > max_sectors) {
- spin_unlock(&dev->lock);
- kfree(area);
- return -EINVAL;
- }
-
- area->begin = *lba = begin;
- area->end = begin + len;
-
- if (prev) /* insert into sorted order */
- list_add(&area->list, &prev->list);
- else
- list_add(&area->list, &dev->area_list);
- spin_unlock(&dev->lock);
-
- return 0;
-}
-EXPORT_SYMBOL(nvm_get_area);
-
-void nvm_put_area(struct nvm_tgt_dev *tgt_dev, sector_t begin)
-{
- struct nvm_dev *dev = tgt_dev->parent;
- struct nvm_area *area;
-
- spin_lock(&dev->lock);
- list_for_each_entry(area, &dev->area_list, list) {
- if (area->begin != begin)
- continue;
-
- list_del(&area->list);
- spin_unlock(&dev->lock);
- kfree(area);
- return;
- }
- spin_unlock(&dev->lock);
-}
-EXPORT_SYMBOL(nvm_put_area);
-
void nvm_end_io(struct nvm_rq *rqd)
{
struct nvm_tgt_dev *tgt_dev = rqd->dev;
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index ba3d7f3..26f7ecc 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -31,27 +31,10 @@
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;
@@ -80,19 +63,6 @@ struct nvme_nvm_identity {
__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;
@@ -139,9 +109,7 @@ 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;
@@ -234,11 +202,9 @@ struct nvme_nvm_bb_tbl {
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) != 16);
@@ -332,62 +298,6 @@ static int nvme_nvm_identity(struct nvm_dev *nvmdev, struct nvm_id *nvm_id)
return ret;
}
-static int nvme_nvm_get_l2p_tbl(struct nvm_dev *nvmdev, u64 slba, u32 nlb,
- nvm_l2p_update_fn *update_l2p, void *priv)
-{
- struct nvme_ns *ns = nvmdev->q->queuedata;
- struct nvme_nvm_command c = {};
- u32 len = queue_max_hw_sectors(ns->ctrl->admin_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->head->ns_id);
- entries = kmalloc(len, GFP_KERNEL);
- if (!entries)
- return -ENOMEM;
-
- while (nlb) {
- u32 cmd_nlb = min(nlb_pr_rq, nlb);
- u64 elba = slba + cmd_nlb;
-
- c.l2p.slba = cpu_to_le64(cmd_slba);
- c.l2p.nlb = cpu_to_le32(cmd_nlb);
-
- ret = nvme_submit_sync_cmd(ns->ctrl->admin_q,
- (struct nvme_command *)&c, entries, len);
- if (ret) {
- dev_err(ns->ctrl->device,
- "L2P table transfer failed (%d)\n", ret);
- ret = -EIO;
- goto out;
- }
-
- if (unlikely(elba > nvmdev->total_secs)) {
- pr_err("nvm: L2P data from device is out of bounds!\n");
- ret = -EINVAL;
- goto out;
- }
-
- /* Transform physical address to target address space */
- nvm_part_to_tgt(nvmdev, entries, cmd_nlb);
-
- 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 nvm_dev *nvmdev, struct ppa_addr ppa,
u8 *blks)
{
@@ -474,10 +384,6 @@ static inline void nvme_nvm_rqtocmd(struct nvm_rq *rqd, struct nvme_ns *ns,
c->ph_rw.metadata = cpu_to_le64(rqd->dma_meta_list);
c->ph_rw.control = cpu_to_le16(rqd->flags);
c->ph_rw.length = cpu_to_le16(rqd->nr_ppas - 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, blk_status_t status)
@@ -597,8 +503,6 @@ static void nvme_nvm_dev_dma_free(void *pool, void *addr,
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,
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 14e274b..97ceb84 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -50,10 +50,7 @@ struct nvm_id;
struct nvm_dev;
struct nvm_tgt_dev;
-typedef int (nvm_l2p_update_fn)(u64, u32, __le64 *, void *);
typedef int (nvm_id_fn)(struct nvm_dev *, struct nvm_id *);
-typedef int (nvm_get_l2p_tbl_fn)(struct nvm_dev *, u64, u32,
- nvm_l2p_update_fn *, void *);
typedef int (nvm_op_bb_tbl_fn)(struct nvm_dev *, struct ppa_addr, u8 *);
typedef int (nvm_op_set_bb_fn)(struct nvm_dev *, struct ppa_addr *, int, int);
typedef int (nvm_submit_io_fn)(struct nvm_dev *, struct nvm_rq *);
@@ -66,7 +63,6 @@ typedef void (nvm_dev_dma_free_fn)(void *, void*, dma_addr_t);
struct nvm_dev_ops {
nvm_id_fn *identity;
- nvm_get_l2p_tbl_fn *get_l2p_tbl;
nvm_op_bb_tbl_fn *get_bb_tbl;
nvm_op_set_bb_fn *set_bb_tbl;
@@ -112,8 +108,6 @@ enum {
NVM_RSP_WARN_HIGHECC = 0x4700,
/* Device opcodes */
- NVM_OP_HBREAD = 0x02,
- NVM_OP_HBWRITE = 0x81,
NVM_OP_PWRITE = 0x91,
NVM_OP_PREAD = 0x92,
NVM_OP_ERASE = 0x90,
@@ -346,36 +340,6 @@ struct nvm_dev {
struct list_head targets;
};
-static inline struct ppa_addr linear_to_generic_addr(struct nvm_geo *geo,
- u64 pba)
-{
- struct ppa_addr l;
- int secs, pgs, blks, luns;
- sector_t ppa = pba;
-
- l.ppa = 0;
-
- div_u64_rem(ppa, geo->sec_per_pg, &secs);
- l.g.sec = secs;
-
- sector_div(ppa, geo->sec_per_pg);
- div_u64_rem(ppa, geo->pgs_per_blk, &pgs);
- l.g.pg = pgs;
-
- sector_div(ppa, geo->pgs_per_blk);
- div_u64_rem(ppa, geo->blks_per_lun, &blks);
- l.g.blk = blks;
-
- sector_div(ppa, geo->blks_per_lun);
- div_u64_rem(ppa, geo->luns_per_chnl, &luns);
- l.g.lun = luns;
-
- sector_div(ppa, geo->luns_per_chnl);
- l.g.ch = ppa;
-
- return l;
-}
-
static inline struct ppa_addr generic_to_dev_addr(struct nvm_tgt_dev *tgt_dev,
struct ppa_addr r)
{
@@ -462,17 +426,10 @@ extern int nvm_set_tgt_bb_tbl(struct nvm_tgt_dev *, struct ppa_addr *,
extern int nvm_max_phys_sects(struct nvm_tgt_dev *);
extern int nvm_submit_io(struct nvm_tgt_dev *, struct nvm_rq *);
extern int nvm_submit_io_sync(struct nvm_tgt_dev *, struct nvm_rq *);
-extern int nvm_erase_sync(struct nvm_tgt_dev *, struct ppa_addr *, int);
-extern int nvm_get_l2p_tbl(struct nvm_tgt_dev *, u64, u32, nvm_l2p_update_fn *,
- void *);
-extern int nvm_get_area(struct nvm_tgt_dev *, sector_t *, sector_t);
-extern void nvm_put_area(struct nvm_tgt_dev *, sector_t);
extern void nvm_end_io(struct nvm_rq *);
extern int nvm_bb_tbl_fold(struct nvm_dev *, u8 *, int);
extern int nvm_get_tgt_bb_tbl(struct nvm_tgt_dev *, struct ppa_addr, u8 *);
-extern void nvm_part_to_tgt(struct nvm_dev *, sector_t *, int);
-
#else /* CONFIG_NVM */
struct nvm_dev_ops;
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 05/25] lightnvm: remove unnecessary field from nvm_rq
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (3 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 04/25] lightnvm: remove hybrid ocssd 1.2 support Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 06/25] lightnvm: remove lower page tables Matias Bjørling
` (20 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Remove the wait filed in nvm_rq. It is not used anymore, as targets rely
on the functionality provided by the LightNVM subsystem when sending
sync I/O.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
include/linux/lightnvm.h | 1 -
1 file changed, 1 deletion(-)
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 97ceb84..07cdb05 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -233,7 +233,6 @@ struct nvm_rq {
void *meta_list;
dma_addr_t dma_meta_list;
- struct completion *wait;
nvm_end_io_fn *end_io;
uint8_t opcode;
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 06/25] lightnvm: remove lower page tables
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (4 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 05/25] lightnvm: remove unnecessary field from nvm_rq Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 07/25] lightnvm: make geometry structures 2.0 ready Matias Bjørling
` (19 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
The lower page table is unused. All page tables reported by 1.2
devices are all reporting a sequential 1:1 page mapping. This is
also not used going forward with the 2.0 revision.
Signed-off-by: Matias Bjørling <m@bjorling.me>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 67 --------------------------------------------
drivers/nvme/host/lightnvm.c | 14 ---------
include/linux/lightnvm.h | 6 ----
3 files changed, 87 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 390d5ef..52059dd 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -751,53 +751,6 @@ int nvm_get_tgt_bb_tbl(struct nvm_tgt_dev *tgt_dev, struct ppa_addr ppa,
}
EXPORT_SYMBOL(nvm_get_tgt_bb_tbl);
-static int nvm_init_slc_tbl(struct nvm_dev *dev, struct nvm_id_group *grp)
-{
- struct nvm_geo *geo = &dev->geo;
- int i;
-
- dev->lps_per_blk = geo->pgs_per_blk;
- dev->lptbl = kcalloc(dev->lps_per_blk, sizeof(int), GFP_KERNEL);
- if (!dev->lptbl)
- return -ENOMEM;
-
- /* Just a linear array */
- for (i = 0; i < dev->lps_per_blk; i++)
- dev->lptbl[i] = i;
-
- return 0;
-}
-
-static int nvm_init_mlc_tbl(struct nvm_dev *dev, struct nvm_id_group *grp)
-{
- int i, p;
- struct nvm_id_lp_mlc *mlc = &grp->lptbl.mlc;
-
- if (!mlc->num_pairs)
- return 0;
-
- dev->lps_per_blk = mlc->num_pairs;
- dev->lptbl = kcalloc(dev->lps_per_blk, sizeof(int), GFP_KERNEL);
- if (!dev->lptbl)
- return -ENOMEM;
-
- /* The lower page table encoding consists of a list of bytes, where each
- * has a lower and an upper half. The first half byte maintains the
- * increment value and every value after is an offset added to the
- * previous incrementation value
- */
- dev->lptbl[0] = mlc->pairs[0] & 0xF;
- for (i = 1; i < dev->lps_per_blk; i++) {
- p = mlc->pairs[i >> 1];
- if (i & 0x1) /* upper */
- dev->lptbl[i] = dev->lptbl[i - 1] + ((p & 0xF0) >> 4);
- else /* lower */
- dev->lptbl[i] = dev->lptbl[i - 1] + (p & 0xF);
- }
-
- return 0;
-}
-
static int nvm_core_init(struct nvm_dev *dev)
{
struct nvm_id *id = &dev->identity;
@@ -846,25 +799,6 @@ static int nvm_core_init(struct nvm_dev *dev)
if (!dev->lun_map)
return -ENOMEM;
- switch (grp->fmtype) {
- case NVM_ID_FMTYPE_SLC:
- if (nvm_init_slc_tbl(dev, grp)) {
- ret = -ENOMEM;
- goto err_fmtype;
- }
- break;
- case NVM_ID_FMTYPE_MLC:
- if (nvm_init_mlc_tbl(dev, grp)) {
- ret = -ENOMEM;
- goto err_fmtype;
- }
- break;
- default:
- pr_err("nvm: flash type not supported\n");
- ret = -EINVAL;
- goto err_fmtype;
- }
-
INIT_LIST_HEAD(&dev->area_list);
INIT_LIST_HEAD(&dev->targets);
mutex_init(&dev->mlock);
@@ -890,7 +824,6 @@ static void nvm_free(struct nvm_dev *dev)
dev->ops->destroy_dma_pool(dev->dma_pool);
nvm_unregister_map(dev);
- kfree(dev->lptbl);
kfree(dev->lun_map);
kfree(dev);
}
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index 26f7ecc..15bf243 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -246,20 +246,6 @@ static int init_grps(struct nvm_id *nvm_id, struct nvme_nvm_id *nvme_nvm_id)
dst->cpar = le16_to_cpu(src->cpar);
- if (dst->fmtype == NVM_ID_FMTYPE_MLC) {
- memcpy(dst->lptbl.id, src->lptbl.id, 8);
- dst->lptbl.mlc.num_pairs =
- le16_to_cpu(src->lptbl.mlc.num_pairs);
-
- if (dst->lptbl.mlc.num_pairs > NVME_NVM_LP_MLC_PAIRS) {
- pr_err("nvm: number of MLC pairs not supported\n");
- return -EINVAL;
- }
-
- memcpy(dst->lptbl.mlc.pairs, src->lptbl.mlc.pairs,
- dst->lptbl.mlc.num_pairs);
- }
-
return 0;
}
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 07cdb05..a5d8e0c 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -174,8 +174,6 @@ struct nvm_id_group {
u32 mpos;
u32 mccap;
u16 cpar;
-
- struct nvm_id_lp_tbl lptbl;
};
struct nvm_addr_format {
@@ -313,10 +311,6 @@ struct nvm_dev {
/* Device information */
struct nvm_geo geo;
- /* lower page table */
- int lps_per_blk;
- int *lptbl;
-
unsigned long total_secs;
unsigned long *lun_map;
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 07/25] lightnvm: make geometry structures 2.0 ready
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (5 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 06/25] lightnvm: remove lower page tables Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 08/25] lightnvm: refactor target type lookup Matias Bjørling
` (18 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Matias Bjørling, Matias Bjørling
From: Matias Bjørling <matias@cnexlabs.com>
Prepare for the 2.0 revision by adapting the geometry
structures to coexist with the 1.2 revision.
Signed-off-by: Matias Bjørling <m@bjorling.me>
Reviewed-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 89 +++++++++++++++++++---------------------
drivers/lightnvm/pblk-core.c | 6 +--
drivers/lightnvm/pblk-init.c | 62 ++++++++++++++--------------
drivers/lightnvm/pblk-recovery.c | 2 +-
drivers/lightnvm/pblk-sysfs.c | 6 +--
drivers/lightnvm/pblk.h | 8 ++--
drivers/nvme/host/lightnvm.c | 79 ++++++++++++++++++++++-------------
include/linux/lightnvm.h | 52 ++++++++++++++---------
8 files changed, 169 insertions(+), 135 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 52059dd..6d6d2c1 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -98,7 +98,7 @@ static void nvm_remove_tgt_dev(struct nvm_tgt_dev *tgt_dev, int clear)
if (clear) {
for (j = 0; j < ch_map->nr_luns; j++) {
int lun = j + lun_offs[j];
- int lunid = (ch * dev->geo.luns_per_chnl) + lun;
+ int lunid = (ch * dev->geo.nr_luns) + lun;
WARN_ON(!test_and_clear_bit(lunid,
dev->lun_map));
@@ -124,10 +124,10 @@ static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
struct ppa_addr *luns;
int nr_luns = lun_end - lun_begin + 1;
int luns_left = nr_luns;
- int nr_chnls = nr_luns / dev->geo.luns_per_chnl;
- int nr_chnls_mod = nr_luns % dev->geo.luns_per_chnl;
- int bch = lun_begin / dev->geo.luns_per_chnl;
- int blun = lun_begin % dev->geo.luns_per_chnl;
+ int nr_chnls = nr_luns / dev->geo.nr_luns;
+ int nr_chnls_mod = nr_luns % dev->geo.nr_luns;
+ int bch = lun_begin / dev->geo.nr_luns;
+ int blun = lun_begin % dev->geo.nr_luns;
int lunid = 0;
int lun_balanced = 1;
int prev_nr_luns;
@@ -148,15 +148,15 @@ static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
if (!luns)
goto err_luns;
- prev_nr_luns = (luns_left > dev->geo.luns_per_chnl) ?
- dev->geo.luns_per_chnl : luns_left;
+ prev_nr_luns = (luns_left > dev->geo.nr_luns) ?
+ dev->geo.nr_luns : luns_left;
for (i = 0; i < nr_chnls; i++) {
struct nvm_ch_map *ch_rmap = &dev_rmap->chnls[i + bch];
int *lun_roffs = ch_rmap->lun_offs;
struct nvm_ch_map *ch_map = &dev_map->chnls[i];
int *lun_offs;
- int luns_in_chnl = (luns_left > dev->geo.luns_per_chnl) ?
- dev->geo.luns_per_chnl : luns_left;
+ int luns_in_chnl = (luns_left > dev->geo.nr_luns) ?
+ dev->geo.nr_luns : luns_left;
if (lun_balanced && prev_nr_luns != luns_in_chnl)
lun_balanced = 0;
@@ -193,8 +193,8 @@ static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
memcpy(&tgt_dev->geo, &dev->geo, sizeof(struct nvm_geo));
/* Target device only owns a portion of the physical device */
tgt_dev->geo.nr_chnls = nr_chnls;
- tgt_dev->geo.nr_luns = nr_luns;
- tgt_dev->geo.luns_per_chnl = (lun_balanced) ? prev_nr_luns : -1;
+ tgt_dev->geo.all_luns = nr_luns;
+ tgt_dev->geo.nr_luns = (lun_balanced) ? prev_nr_luns : -1;
tgt_dev->total_secs = nr_luns * tgt_dev->geo.sec_per_lun;
tgt_dev->q = dev->q;
tgt_dev->map = dev_map;
@@ -414,7 +414,7 @@ static int nvm_register_map(struct nvm_dev *dev)
for (i = 0; i < dev->geo.nr_chnls; i++) {
struct nvm_ch_map *ch_rmap;
int *lun_roffs;
- int luns_in_chnl = dev->geo.luns_per_chnl;
+ int luns_in_chnl = dev->geo.nr_luns;
ch_rmap = &rmap->chnls[i];
@@ -717,10 +717,10 @@ int nvm_bb_tbl_fold(struct nvm_dev *dev, u8 *blks, int nr_blks)
struct nvm_geo *geo = &dev->geo;
int blk, offset, pl, blktype;
- if (nr_blks != geo->blks_per_lun * geo->plane_mode)
+ if (nr_blks != geo->nr_chks * geo->plane_mode)
return -EINVAL;
- for (blk = 0; blk < geo->blks_per_lun; blk++) {
+ for (blk = 0; blk < geo->nr_chks; blk++) {
offset = blk * geo->plane_mode;
blktype = blks[offset];
@@ -736,7 +736,7 @@ int nvm_bb_tbl_fold(struct nvm_dev *dev, u8 *blks, int nr_blks)
blks[blk] = blktype;
}
- return geo->blks_per_lun;
+ return geo->nr_chks;
}
EXPORT_SYMBOL(nvm_bb_tbl_fold);
@@ -758,43 +758,40 @@ static int nvm_core_init(struct nvm_dev *dev)
struct nvm_geo *geo = &dev->geo;
int ret;
+ memcpy(&geo->ppaf, &id->ppaf, sizeof(struct nvm_addr_format));
+
+ if (grp->mtype != 0) {
+ pr_err("nvm: memory type not supported\n");
+ return -EINVAL;
+ }
+
/* Whole device values */
geo->nr_chnls = grp->num_ch;
- geo->luns_per_chnl = grp->num_lun;
+ geo->nr_luns = grp->num_lun;
- /* Generic device values */
- geo->pgs_per_blk = grp->num_pg;
- geo->blks_per_lun = grp->num_blk;
- geo->nr_planes = grp->num_pln;
- geo->fpg_size = grp->fpg_sz;
- geo->pfpg_size = grp->fpg_sz * grp->num_pln;
+ /* Generic device geometry values */
+ geo->ws_min = grp->ws_min;
+ geo->ws_opt = grp->ws_opt;
+ geo->ws_seq = grp->ws_seq;
+ geo->ws_per_chk = grp->ws_per_chk;
+ geo->nr_chks = grp->num_chk;
geo->sec_size = grp->csecs;
geo->oob_size = grp->sos;
- geo->sec_per_pg = grp->fpg_sz / grp->csecs;
geo->mccap = grp->mccap;
- memcpy(&geo->ppaf, &id->ppaf, sizeof(struct nvm_addr_format));
-
- geo->plane_mode = NVM_PLANE_SINGLE;
geo->max_rq_size = dev->ops->max_phys_sect * geo->sec_size;
- if (grp->mpos & 0x020202)
- geo->plane_mode = NVM_PLANE_DOUBLE;
- if (grp->mpos & 0x040404)
- geo->plane_mode = NVM_PLANE_QUAD;
+ geo->sec_per_chk = grp->clba;
+ geo->sec_per_lun = geo->sec_per_chk * geo->nr_chks;
+ geo->all_luns = geo->nr_luns * geo->nr_chnls;
- if (grp->mtype != 0) {
- pr_err("nvm: memory type not supported\n");
- return -EINVAL;
- }
-
- /* calculated values */
+ /* 1.2 spec device geometry values */
+ geo->plane_mode = 1 << geo->ws_seq;
+ geo->nr_planes = geo->ws_opt / geo->ws_min;
+ geo->sec_per_pg = geo->ws_min;
geo->sec_per_pl = geo->sec_per_pg * geo->nr_planes;
- geo->sec_per_blk = geo->sec_per_pl * geo->pgs_per_blk;
- geo->sec_per_lun = geo->sec_per_blk * geo->blks_per_lun;
- geo->nr_luns = geo->luns_per_chnl * geo->nr_chnls;
- dev->total_secs = geo->nr_luns * geo->sec_per_lun;
- dev->lun_map = kcalloc(BITS_TO_LONGS(geo->nr_luns),
+ dev->total_secs = geo->all_luns * geo->sec_per_lun;
+ dev->lun_map = kcalloc(BITS_TO_LONGS(geo->all_luns),
sizeof(unsigned long), GFP_KERNEL);
if (!dev->lun_map)
return -ENOMEM;
@@ -854,8 +851,8 @@ static int nvm_init(struct nvm_dev *dev)
pr_info("nvm: registered %s [%u/%u/%u/%u/%u/%u]\n",
dev->name, geo->sec_per_pg, geo->nr_planes,
- geo->pgs_per_blk, geo->blks_per_lun,
- geo->nr_luns, geo->nr_chnls);
+ geo->ws_per_chk, geo->nr_chks,
+ geo->all_luns, geo->nr_chnls);
return 0;
err:
pr_err("nvm: failed to initialize nvm\n");
@@ -946,12 +943,12 @@ static int __nvm_configure_create(struct nvm_ioctl_create *create)
if (s->lun_begin == -1 && s->lun_end == -1) {
s->lun_begin = 0;
- s->lun_end = dev->geo.nr_luns - 1;
+ s->lun_end = dev->geo.all_luns - 1;
}
- if (s->lun_begin > s->lun_end || s->lun_end >= dev->geo.nr_luns) {
+ if (s->lun_begin > s->lun_end || s->lun_end >= dev->geo.all_luns) {
pr_err("nvm: lun out of bound (%u:%u > %u)\n",
- s->lun_begin, s->lun_end, dev->geo.nr_luns - 1);
+ s->lun_begin, s->lun_end, dev->geo.all_luns - 1);
return -EINVAL;
}
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 76516ee..0849046 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -979,7 +979,7 @@ static int pblk_line_init_metadata(struct pblk *pblk, struct pblk_line *line,
/* Start metadata */
smeta_buf->seq_nr = cpu_to_le64(line->seq_nr);
- smeta_buf->window_wr_lun = cpu_to_le32(geo->nr_luns);
+ smeta_buf->window_wr_lun = cpu_to_le32(geo->all_luns);
/* Fill metadata among lines */
if (cur) {
@@ -1032,7 +1032,7 @@ static int pblk_line_init_bb(struct pblk *pblk, struct pblk_line *line,
lm->sec_per_line);
bitmap_or(line->map_bitmap, line->map_bitmap, l_mg->bb_aux,
lm->sec_per_line);
- line->sec_in_line -= geo->sec_per_blk;
+ line->sec_in_line -= geo->sec_per_chk;
if (bit >= lm->emeta_bb)
nr_bb++;
}
@@ -1746,7 +1746,7 @@ void pblk_up_rq(struct pblk *pblk, struct ppa_addr *ppa_list, int nr_ppas,
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
struct pblk_lun *rlun;
- int nr_luns = geo->nr_luns;
+ int nr_luns = geo->all_luns;
int bit = -1;
while ((bit = find_next_bit(lun_bitmap, nr_luns, bit + 1)) < nr_luns) {
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 695826a..d13bb51 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -169,8 +169,8 @@ static int pblk_set_ppaf(struct pblk *pblk)
}
ppaf.ch_len = power_len;
- power_len = get_count_order(geo->luns_per_chnl);
- if (1 << power_len != geo->luns_per_chnl) {
+ power_len = get_count_order(geo->nr_luns);
+ if (1 << power_len != geo->nr_luns) {
pr_err("pblk: supports only power-of-two LUN config.\n");
return -EINVAL;
}
@@ -254,7 +254,7 @@ static int pblk_core_init(struct pblk *pblk)
struct nvm_geo *geo = &dev->geo;
pblk->pgs_in_buffer = NVM_MEM_PAGE_WRITE * geo->sec_per_pg *
- geo->nr_planes * geo->nr_luns;
+ geo->nr_planes * geo->all_luns;
if (pblk_init_global_caches(pblk))
return -ENOMEM;
@@ -270,21 +270,22 @@ static int pblk_core_init(struct pblk *pblk)
if (!pblk->gen_ws_pool)
goto free_page_bio_pool;
- pblk->rec_pool = mempool_create_slab_pool(geo->nr_luns, pblk_rec_cache);
+ pblk->rec_pool = mempool_create_slab_pool(geo->all_luns,
+ pblk_rec_cache);
if (!pblk->rec_pool)
goto free_gen_ws_pool;
- pblk->r_rq_pool = mempool_create_slab_pool(geo->nr_luns,
+ pblk->r_rq_pool = mempool_create_slab_pool(geo->all_luns,
pblk_g_rq_cache);
if (!pblk->r_rq_pool)
goto free_rec_pool;
- pblk->e_rq_pool = mempool_create_slab_pool(geo->nr_luns,
+ pblk->e_rq_pool = mempool_create_slab_pool(geo->all_luns,
pblk_g_rq_cache);
if (!pblk->e_rq_pool)
goto free_r_rq_pool;
- pblk->w_rq_pool = mempool_create_slab_pool(geo->nr_luns,
+ pblk->w_rq_pool = mempool_create_slab_pool(geo->all_luns,
pblk_w_rq_cache);
if (!pblk->w_rq_pool)
goto free_e_rq_pool;
@@ -409,7 +410,7 @@ static int pblk_bb_discovery(struct nvm_tgt_dev *dev, struct pblk_lun *rlun)
u8 *blks;
int nr_blks, ret;
- nr_blks = geo->blks_per_lun * geo->plane_mode;
+ nr_blks = geo->nr_chks * geo->plane_mode;
blks = kmalloc(nr_blks, GFP_KERNEL);
if (!blks)
return -ENOMEM;
@@ -482,20 +483,21 @@ static int pblk_luns_init(struct pblk *pblk, struct ppa_addr *luns)
int i, ret;
/* TODO: Implement unbalanced LUN support */
- if (geo->luns_per_chnl < 0) {
+ if (geo->nr_luns < 0) {
pr_err("pblk: unbalanced LUN config.\n");
return -EINVAL;
}
- pblk->luns = kcalloc(geo->nr_luns, sizeof(struct pblk_lun), GFP_KERNEL);
+ pblk->luns = kcalloc(geo->all_luns, sizeof(struct pblk_lun),
+ GFP_KERNEL);
if (!pblk->luns)
return -ENOMEM;
- for (i = 0; i < geo->nr_luns; i++) {
+ for (i = 0; i < geo->all_luns; i++) {
/* Stripe across channels */
int ch = i % geo->nr_chnls;
int lun_raw = i / geo->nr_chnls;
- int lunid = lun_raw + ch * geo->luns_per_chnl;
+ int lunid = lun_raw + ch * geo->nr_luns;
rlun = &pblk->luns[i];
rlun->bppa = luns[lunid];
@@ -590,8 +592,8 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks)
* on user capacity consider only provisioned blocks
*/
pblk->rl.total_blocks = nr_free_blks;
- pblk->rl.nr_secs = nr_free_blks * geo->sec_per_blk;
- pblk->capacity = provisioned * geo->sec_per_blk;
+ pblk->rl.nr_secs = nr_free_blks * geo->sec_per_chk;
+ pblk->capacity = provisioned * geo->sec_per_chk;
atomic_set(&pblk->rl.free_blocks, nr_free_blks);
}
@@ -683,7 +685,7 @@ static int pblk_lines_init(struct pblk *pblk)
int i, ret;
pblk->min_write_pgs = geo->sec_per_pl * (geo->sec_size / PAGE_SIZE);
- max_write_ppas = pblk->min_write_pgs * geo->nr_luns;
+ max_write_ppas = pblk->min_write_pgs * geo->all_luns;
pblk->max_write_pgs = (max_write_ppas < nvm_max_phys_sects(dev)) ?
max_write_ppas : nvm_max_phys_sects(dev);
pblk_set_sec_per_write(pblk, pblk->min_write_pgs);
@@ -693,26 +695,26 @@ static int pblk_lines_init(struct pblk *pblk)
return -EINVAL;
}
- div_u64_rem(geo->sec_per_blk, pblk->min_write_pgs, &mod);
+ div_u64_rem(geo->sec_per_chk, pblk->min_write_pgs, &mod);
if (mod) {
pr_err("pblk: bad configuration of sectors/pages\n");
return -EINVAL;
}
- l_mg->nr_lines = geo->blks_per_lun;
+ l_mg->nr_lines = geo->nr_chks;
l_mg->log_line = l_mg->data_line = NULL;
l_mg->l_seq_nr = l_mg->d_seq_nr = 0;
l_mg->nr_free_lines = 0;
bitmap_zero(&l_mg->meta_bitmap, PBLK_DATA_LINES);
- lm->sec_per_line = geo->sec_per_blk * geo->nr_luns;
- lm->blk_per_line = geo->nr_luns;
- lm->blk_bitmap_len = BITS_TO_LONGS(geo->nr_luns) * sizeof(long);
+ lm->sec_per_line = geo->sec_per_chk * geo->all_luns;
+ lm->blk_per_line = geo->all_luns;
+ lm->blk_bitmap_len = BITS_TO_LONGS(geo->all_luns) * sizeof(long);
lm->sec_bitmap_len = BITS_TO_LONGS(lm->sec_per_line) * sizeof(long);
- lm->lun_bitmap_len = BITS_TO_LONGS(geo->nr_luns) * sizeof(long);
+ lm->lun_bitmap_len = BITS_TO_LONGS(geo->all_luns) * sizeof(long);
lm->mid_thrs = lm->sec_per_line / 2;
lm->high_thrs = lm->sec_per_line / 4;
- lm->meta_distance = (geo->nr_luns / 2) * pblk->min_write_pgs;
+ lm->meta_distance = (geo->all_luns / 2) * pblk->min_write_pgs;
/* Calculate necessary pages for smeta. See comment over struct
* line_smeta definition
@@ -742,12 +744,12 @@ static int pblk_lines_init(struct pblk *pblk)
goto add_emeta_page;
}
- lm->emeta_bb = geo->nr_luns > i ? geo->nr_luns - i : 0;
+ lm->emeta_bb = geo->all_luns > i ? geo->all_luns - i : 0;
lm->min_blk_line = 1;
- if (geo->nr_luns > 1)
+ if (geo->all_luns > 1)
lm->min_blk_line += DIV_ROUND_UP(lm->smeta_sec +
- lm->emeta_sec[0], geo->sec_per_blk);
+ lm->emeta_sec[0], geo->sec_per_chk);
if (lm->min_blk_line > lm->blk_per_line) {
pr_err("pblk: config. not supported. Min. LUN in line:%d\n",
@@ -772,7 +774,7 @@ static int pblk_lines_init(struct pblk *pblk)
goto fail_free_bb_template;
}
- bb_distance = (geo->nr_luns) * geo->sec_per_pl;
+ bb_distance = (geo->all_luns) * geo->sec_per_pl;
for (i = 0; i < lm->sec_per_line; i += bb_distance)
bitmap_set(l_mg->bb_template, i, geo->sec_per_pl);
@@ -844,7 +846,7 @@ static int pblk_lines_init(struct pblk *pblk)
pblk_set_provision(pblk, nr_free_blks);
/* Cleanup per-LUN bad block lists - managed within lines on run-time */
- for (i = 0; i < geo->nr_luns; i++)
+ for (i = 0; i < geo->all_luns; i++)
kfree(pblk->luns[i].bb_list);
return 0;
@@ -858,7 +860,7 @@ static int pblk_lines_init(struct pblk *pblk)
fail_free_meta:
pblk_line_meta_free(pblk);
fail:
- for (i = 0; i < geo->nr_luns; i++)
+ for (i = 0; i < geo->all_luns; i++)
kfree(pblk->luns[i].bb_list);
return ret;
@@ -1041,13 +1043,13 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
blk_queue_write_cache(tqueue, true, false);
- tqueue->limits.discard_granularity = geo->pgs_per_blk * geo->pfpg_size;
+ tqueue->limits.discard_granularity = geo->sec_per_chk * geo->sec_size;
tqueue->limits.discard_alignment = 0;
blk_queue_max_discard_sectors(tqueue, UINT_MAX >> 9);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, tqueue);
pr_info("pblk init: luns:%u, lines:%d, secs:%llu, buf entries:%u\n",
- geo->nr_luns, pblk->l_mg.nr_lines,
+ geo->all_luns, pblk->l_mg.nr_lines,
(unsigned long long)pblk->rl.nr_secs,
pblk->rwb.nr_entries);
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index eadb3eb..ceec12d 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -188,7 +188,7 @@ static int pblk_calc_sec_in_line(struct pblk *pblk, struct pblk_line *line)
int nr_bb = bitmap_weight(line->blk_bitmap, lm->blk_per_line);
return lm->sec_per_line - lm->smeta_sec - lm->emeta_sec[0] -
- nr_bb * geo->sec_per_blk;
+ nr_bb * geo->sec_per_chk;
}
struct pblk_recov_alloc {
diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index cd49e88..5cee2ac 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -28,7 +28,7 @@ static ssize_t pblk_sysfs_luns_show(struct pblk *pblk, char *page)
ssize_t sz = 0;
int i;
- for (i = 0; i < geo->nr_luns; i++) {
+ for (i = 0; i < geo->all_luns; i++) {
int active = 1;
rlun = &pblk->luns[i];
@@ -238,7 +238,7 @@ static ssize_t pblk_sysfs_lines(struct pblk *pblk, char *page)
sz = snprintf(page, PAGE_SIZE - sz,
"line: nluns:%d, nblks:%d, nsecs:%d\n",
- geo->nr_luns, lm->blk_per_line, lm->sec_per_line);
+ geo->all_luns, lm->blk_per_line, lm->sec_per_line);
sz += snprintf(page + sz, PAGE_SIZE - sz,
"lines:d:%d,l:%d-f:%d,m:%d/%d,c:%d,b:%d,co:%d(d:%d,l:%d)t:%d\n",
@@ -287,7 +287,7 @@ static ssize_t pblk_sysfs_lines_info(struct pblk *pblk, char *page)
"blk_line:%d, sec_line:%d, sec_blk:%d\n",
lm->blk_per_line,
lm->sec_per_line,
- geo->sec_per_blk);
+ geo->sec_per_chk);
return sz;
}
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 59a64d4..c150728 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -907,7 +907,7 @@ static inline int pblk_pad_distance(struct pblk *pblk)
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
- return NVM_MEM_PAGE_WRITE * geo->nr_luns * geo->sec_per_pl;
+ return NVM_MEM_PAGE_WRITE * geo->all_luns * geo->sec_per_pl;
}
static inline int pblk_dev_ppa_to_line(struct ppa_addr p)
@@ -1212,10 +1212,10 @@ static inline int pblk_boundary_ppa_checks(struct nvm_tgt_dev *tgt_dev,
if (!ppa->c.is_cached &&
ppa->g.ch < geo->nr_chnls &&
- ppa->g.lun < geo->luns_per_chnl &&
+ ppa->g.lun < geo->nr_luns &&
ppa->g.pl < geo->nr_planes &&
- ppa->g.blk < geo->blks_per_lun &&
- ppa->g.pg < geo->pgs_per_blk &&
+ ppa->g.blk < geo->nr_chks &&
+ ppa->g.pg < geo->ws_per_chk &&
ppa->g.sec < geo->sec_per_pg)
continue;
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index 15bf243..50ef71ee 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -135,7 +135,7 @@ struct nvme_nvm_id_group {
__u8 num_lun;
__u8 num_pln;
__u8 rsvd1;
- __le16 num_blk;
+ __le16 num_chk;
__le16 num_pg;
__le16 fpg_sz;
__le16 csecs;
@@ -215,36 +215,57 @@ static inline void _nvme_nvm_check_size(void)
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;
+ struct nvm_id_group *grp;
+ int sec_per_pg, sec_per_pl, pg_per_blk;
if (nvme_nvm_id->cgrps != 1)
return -EINVAL;
src = &nvme_nvm_id->groups[0];
- dst = &nvm_id->grp;
-
- 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);
+ grp = &nvm_id->grp;
+
+ grp->mtype = src->mtype;
+ grp->fmtype = src->fmtype;
+
+ grp->num_ch = src->num_ch;
+ grp->num_lun = src->num_lun;
+
+ grp->num_chk = le16_to_cpu(src->num_chk);
+ grp->csecs = le16_to_cpu(src->csecs);
+ grp->sos = le16_to_cpu(src->sos);
+
+ pg_per_blk = le16_to_cpu(src->num_pg);
+ sec_per_pg = le16_to_cpu(src->fpg_sz) / grp->csecs;
+ sec_per_pl = sec_per_pg * src->num_pln;
+ grp->clba = sec_per_pl * pg_per_blk;
+ grp->ws_per_chk = pg_per_blk;
+
+ grp->mpos = le32_to_cpu(src->mpos);
+ grp->cpar = le16_to_cpu(src->cpar);
+ grp->mccap = le32_to_cpu(src->mccap);
+
+ grp->ws_opt = grp->ws_min = sec_per_pg;
+ grp->ws_seq = NVM_IO_SNGL_ACCESS;
+
+ if (grp->mpos & 0x020202) {
+ grp->ws_seq = NVM_IO_DUAL_ACCESS;
+ grp->ws_opt <<= 1;
+ } else if (grp->mpos & 0x040404) {
+ grp->ws_seq = NVM_IO_QUAD_ACCESS;
+ grp->ws_opt <<= 2;
+ }
+
+ grp->trdt = le32_to_cpu(src->trdt);
+ grp->trdm = le32_to_cpu(src->trdm);
+ grp->tprt = le32_to_cpu(src->tprt);
+ grp->tprm = le32_to_cpu(src->tprm);
+ grp->tbet = le32_to_cpu(src->tbet);
+ grp->tbem = le32_to_cpu(src->tbem);
+
+ /* 1.2 compatibility */
+ grp->num_pln = src->num_pln;
+ grp->num_pg = le16_to_cpu(src->num_pg);
+ grp->fpg_sz = le16_to_cpu(src->fpg_sz);
return 0;
}
@@ -293,7 +314,7 @@ static int nvme_nvm_get_bb_tbl(struct nvm_dev *nvmdev, struct ppa_addr ppa,
struct nvme_ctrl *ctrl = ns->ctrl;
struct nvme_nvm_command c = {};
struct nvme_nvm_bb_tbl *bb_tbl;
- int nr_blks = geo->blks_per_lun * geo->plane_mode;
+ int nr_blks = geo->nr_chks * geo->plane_mode;
int tblsz = sizeof(struct nvme_nvm_bb_tbl) + nr_blks;
int ret = 0;
@@ -334,7 +355,7 @@ static int nvme_nvm_get_bb_tbl(struct nvm_dev *nvmdev, struct ppa_addr ppa,
goto out;
}
- memcpy(blks, bb_tbl->blk, geo->blks_per_lun * geo->plane_mode);
+ memcpy(blks, bb_tbl->blk, geo->nr_chks * geo->plane_mode);
out:
kfree(bb_tbl);
return ret;
@@ -773,7 +794,7 @@ static ssize_t nvm_dev_attr_show(struct device *dev,
} else if (strcmp(attr->name, "num_planes") == 0) {
return scnprintf(page, PAGE_SIZE, "%u\n", grp->num_pln);
} else if (strcmp(attr->name, "num_blocks") == 0) { /* u16 */
- return scnprintf(page, PAGE_SIZE, "%u\n", grp->num_blk);
+ return scnprintf(page, PAGE_SIZE, "%u\n", grp->num_chk);
} else if (strcmp(attr->name, "num_pages") == 0) {
return scnprintf(page, PAGE_SIZE, "%u\n", grp->num_pg);
} else if (strcmp(attr->name, "page_size") == 0) {
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index a5d8e0c..8e43bfe 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -159,12 +159,16 @@ struct nvm_id_group {
u8 fmtype;
u8 num_ch;
u8 num_lun;
- u8 num_pln;
- u16 num_blk;
- u16 num_pg;
- u16 fpg_sz;
+ u16 num_chk;
+ u16 clba;
u16 csecs;
u16 sos;
+
+ u16 ws_min;
+ u16 ws_opt;
+ u16 ws_seq;
+ u16 ws_per_chk;
+
u32 trdt;
u32 trdm;
u32 tprt;
@@ -174,6 +178,11 @@ struct nvm_id_group {
u32 mpos;
u32 mccap;
u16 cpar;
+
+ /* 1.2 compatibility */
+ u8 num_pln;
+ u16 num_pg;
+ u16 fpg_sz;
};
struct nvm_addr_format {
@@ -259,31 +268,36 @@ enum {
NVM_BLK_ST_BAD = 0x8, /* Bad block */
};
+
/* Device generic information */
struct nvm_geo {
+ /* generic geometry */
int nr_chnls;
- int nr_luns;
- int luns_per_chnl; /* -1 if channels are not symmetric */
- int nr_planes;
- int sec_per_pg; /* only sectors for a single page */
- int pgs_per_blk;
- int blks_per_lun;
- int fpg_size;
- int pfpg_size; /* size of buffer if all pages are to be read */
+ int all_luns; /* across channels */
+ int nr_luns; /* per channel */
+ int nr_chks; /* per lun */
+
int sec_size;
int oob_size;
int mccap;
- struct nvm_addr_format ppaf;
- /* Calculated/Cached values. These do not reflect the actual usable
- * blocks at run-time.
- */
+ int sec_per_chk;
+ int sec_per_lun;
+
+ int ws_min;
+ int ws_opt;
+ int ws_seq;
+ int ws_per_chk;
+
int max_rq_size;
+
+ struct nvm_addr_format ppaf;
+
+ /* Legacy 1.2 specific geometry */
int plane_mode; /* drive device in single, double or quad mode */
-
+ int nr_planes;
+ int sec_per_pg; /* only sectors for a single page */
int sec_per_pl; /* all sectors across planes */
- int sec_per_blk;
- int sec_per_lun;
};
/* sub-device structure */
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 08/25] lightnvm: refactor target type lookup
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (6 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 07/25] lightnvm: make geometry structures 2.0 ready Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 09/25] lightnvm: guarantee target unique name across devs Matias Bjørling
` (17 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Refactor target type lookup to use/not use locks explicitly instead of
using a hidden parameter to make the function locking.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 30 +++++++++++++++++-------------
1 file changed, 17 insertions(+), 13 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 6d6d2c1..5c2d0f3 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -220,21 +220,25 @@ static const struct block_device_operations nvm_fops = {
.owner = THIS_MODULE,
};
-static struct nvm_tgt_type *nvm_find_target_type(const char *name, int lock)
+static struct nvm_tgt_type *__nvm_find_target_type(const char *name)
{
- struct nvm_tgt_type *tmp, *tt = NULL;
+ struct nvm_tgt_type *tt;
- if (lock)
- down_write(&nvm_tgtt_lock);
+ list_for_each_entry(tt, &nvm_tgt_types, list)
+ if (!strcmp(name, tt->name))
+ return tt;
- list_for_each_entry(tmp, &nvm_tgt_types, list)
- if (!strcmp(name, tmp->name)) {
- tt = tmp;
- break;
- }
+ return NULL;
+}
+
+static struct nvm_tgt_type *nvm_find_target_type(const char *name)
+{
+ struct nvm_tgt_type *tt;
+
+ down_write(&nvm_tgtt_lock);
+ tt = __nvm_find_target_type(name);
+ up_write(&nvm_tgtt_lock);
- if (lock)
- up_write(&nvm_tgtt_lock);
return tt;
}
@@ -249,7 +253,7 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
void *targetdata;
int ret;
- tt = nvm_find_target_type(create->tgttype, 1);
+ tt = nvm_find_target_type(create->tgttype);
if (!tt) {
pr_err("nvm: target type %s not found\n", create->tgttype);
return -EINVAL;
@@ -523,7 +527,7 @@ int nvm_register_tgt_type(struct nvm_tgt_type *tt)
int ret = 0;
down_write(&nvm_tgtt_lock);
- if (nvm_find_target_type(tt->name, 0))
+ if (__nvm_find_target_type(tt->name))
ret = -EEXIST;
else
list_add(&tt->list, &nvm_tgt_types);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 09/25] lightnvm: guarantee target unique name across devs.
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (7 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 08/25] lightnvm: refactor target type lookup Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 10/25] lightnvm: pblk: compress and reorder helper functions Matias Bjørling
` (16 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Until now, target unique naming is only guaranteed per device. This is
ok from a lightnvm perspective, but not from a sysfs one, since groups
will collide regardless of the underlying device.
Check that names are unique across all lightnvm-capable devices.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 33 +++++++++++++++++++++++++++------
1 file changed, 27 insertions(+), 6 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 5c2d0f3..d5f231c 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -56,6 +56,30 @@ static struct nvm_target *nvm_find_target(struct nvm_dev *dev, const char *name)
return NULL;
}
+static bool nvm_target_exists(const char *name)
+{
+ struct nvm_dev *dev;
+ struct nvm_target *tgt;
+ bool ret = false;
+
+ down_write(&nvm_lock);
+ list_for_each_entry(dev, &nvm_devices, devices) {
+ mutex_lock(&dev->mlock);
+ list_for_each_entry(tgt, &dev->targets, list) {
+ if (!strcmp(name, tgt->disk->disk_name)) {
+ ret = true;
+ mutex_unlock(&dev->mlock);
+ goto out;
+ }
+ }
+ mutex_unlock(&dev->mlock);
+ }
+
+out:
+ up_write(&nvm_lock);
+ return ret;
+}
+
static int nvm_reserve_luns(struct nvm_dev *dev, int lun_begin, int lun_end)
{
int i;
@@ -259,14 +283,11 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
return -EINVAL;
}
- mutex_lock(&dev->mlock);
- t = nvm_find_target(dev, create->tgtname);
- if (t) {
- pr_err("nvm: target name already exists.\n");
- mutex_unlock(&dev->mlock);
+ if (nvm_target_exists(create->tgtname)) {
+ pr_err("nvm: target name already exists (%s)\n",
+ create->tgtname);
return -EINVAL;
}
- mutex_unlock(&dev->mlock);
ret = nvm_reserve_luns(dev, s->lun_begin, s->lun_end);
if (ret)
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 10/25] lightnvm: pblk: compress and reorder helper functions
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (8 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 09/25] lightnvm: guarantee target unique name across devs Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 11/25] lightnvm: pblk: remove pblk_for_each_lun helper Matias Bjørling
` (15 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Through time, we have generated some redundant helper functions.
Refactor them to eliminate redundant and unnecessary code. Also, reorder
them to improve readability
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-core.c | 24 ++++----
drivers/lightnvm/pblk-rb.c | 2 +-
drivers/lightnvm/pblk-read.c | 4 +-
drivers/lightnvm/pblk-recovery.c | 20 +++----
drivers/lightnvm/pblk.h | 121 +++++++++++++++------------------------
5 files changed, 72 insertions(+), 99 deletions(-)
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 0849046..54d0cef 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -32,8 +32,8 @@ static void pblk_line_mark_bb(struct work_struct *work)
struct pblk_line *line;
int pos;
- line = &pblk->lines[pblk_dev_ppa_to_line(*ppa)];
- pos = pblk_dev_ppa_to_pos(&dev->geo, *ppa);
+ line = &pblk->lines[pblk_ppa_to_line(*ppa)];
+ pos = pblk_ppa_to_pos(&dev->geo, *ppa);
pr_err("pblk: failed to mark bb, line:%d, pos:%d\n",
line->id, pos);
@@ -48,7 +48,7 @@ static void pblk_mark_bb(struct pblk *pblk, struct pblk_line *line,
{
struct nvm_tgt_dev *dev = pblk->dev;
struct nvm_geo *geo = &dev->geo;
- int pos = pblk_dev_ppa_to_pos(geo, *ppa);
+ int pos = pblk_ppa_to_pos(geo, *ppa);
pr_debug("pblk: erase failed: line:%d, pos:%d\n", line->id, pos);
atomic_long_inc(&pblk->erase_failed);
@@ -66,7 +66,7 @@ static void __pblk_end_io_erase(struct pblk *pblk, struct nvm_rq *rqd)
{
struct pblk_line *line;
- line = &pblk->lines[pblk_dev_ppa_to_line(rqd->ppa_addr)];
+ line = &pblk->lines[pblk_ppa_to_line(rqd->ppa_addr)];
atomic_dec(&line->left_seblks);
if (rqd->error) {
@@ -144,7 +144,7 @@ void pblk_map_invalidate(struct pblk *pblk, struct ppa_addr ppa)
BUG_ON(pblk_ppa_empty(ppa));
#endif
- line_id = pblk_tgt_ppa_to_line(ppa);
+ line_id = pblk_ppa_to_line(ppa);
line = &pblk->lines[line_id];
paddr = pblk_dev_ppa_to_line_addr(pblk, ppa);
@@ -650,7 +650,7 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
} else {
for (i = 0; i < rqd.nr_ppas; ) {
struct ppa_addr ppa = addr_to_gen_ppa(pblk, paddr, id);
- int pos = pblk_dev_ppa_to_pos(geo, ppa);
+ int pos = pblk_ppa_to_pos(geo, ppa);
int read_type = PBLK_READ_RANDOM;
if (pblk_io_aligned(pblk, rq_ppas))
@@ -668,7 +668,7 @@ static int pblk_line_submit_emeta_io(struct pblk *pblk, struct pblk_line *line,
}
ppa = addr_to_gen_ppa(pblk, paddr, id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
}
if (pblk_boundary_paddr_checks(pblk, paddr + min)) {
@@ -854,8 +854,8 @@ static int pblk_blk_erase_sync(struct pblk *pblk, struct ppa_addr ppa)
struct nvm_geo *geo = &dev->geo;
pr_err("pblk: could not sync erase line:%d,blk:%d\n",
- pblk_dev_ppa_to_line(ppa),
- pblk_dev_ppa_to_pos(geo, ppa));
+ pblk_ppa_to_line(ppa),
+ pblk_ppa_to_pos(geo, ppa));
rqd.error = ret;
goto out;
@@ -1561,8 +1561,8 @@ int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr ppa)
struct nvm_geo *geo = &dev->geo;
pr_err("pblk: could not async erase line:%d,blk:%d\n",
- pblk_dev_ppa_to_line(ppa),
- pblk_dev_ppa_to_pos(geo, ppa));
+ pblk_ppa_to_line(ppa),
+ pblk_ppa_to_pos(geo, ppa));
}
return err;
@@ -1884,7 +1884,7 @@ void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
/* If the L2P entry maps to a line, the reference is valid */
if (!pblk_ppa_empty(ppa) && !pblk_addr_in_cache(ppa)) {
- int line_id = pblk_dev_ppa_to_line(ppa);
+ int line_id = pblk_ppa_to_line(ppa);
struct pblk_line *line = &pblk->lines[line_id];
kref_get(&line->ref);
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index b8f78e4..62db408 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -226,7 +226,7 @@ static int __pblk_rb_update_l2p(struct pblk_rb *rb, unsigned int to_update)
pblk_update_map_dev(pblk, w_ctx->lba, w_ctx->ppa,
entry->cacheline);
- line = &pblk->lines[pblk_tgt_ppa_to_line(w_ctx->ppa)];
+ line = &pblk->lines[pblk_ppa_to_line(w_ctx->ppa)];
kref_put(&line->ref, pblk_line_put);
clean_wctx(w_ctx);
rb->l2p_update = (rb->l2p_update + 1) & (rb->nr_entries - 1);
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index ca79d8f..0fe0c04 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -141,7 +141,7 @@ static void pblk_read_put_rqd_kref(struct pblk *pblk, struct nvm_rq *rqd)
struct ppa_addr ppa = ppa_list[i];
struct pblk_line *line;
- line = &pblk->lines[pblk_dev_ppa_to_line(ppa)];
+ line = &pblk->lines[pblk_ppa_to_line(ppa)];
kref_put(&line->ref, pblk_line_put_wq);
}
}
@@ -270,7 +270,7 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
i = 0;
hole = find_first_zero_bit(read_bitmap, nr_secs);
do {
- int line_id = pblk_dev_ppa_to_line(rqd->ppa_list[i]);
+ int line_id = pblk_ppa_to_line(rqd->ppa_list[i]);
struct pblk_line *line = &pblk->lines[line_id];
kref_put(&line->ref, pblk_line_put);
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index ceec12d..1b272ae 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -149,7 +149,7 @@ static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
struct ppa_addr ppa;
int pos;
- ppa = addr_to_pblk_ppa(pblk, i, line->id);
+ ppa = addr_to_gen_ppa(pblk, i, line->id);
pos = pblk_ppa_to_pos(geo, ppa);
/* Do not update bad blocks */
@@ -263,12 +263,12 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
int pos;
ppa = addr_to_gen_ppa(pblk, r_ptr_int, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
while (test_bit(pos, line->blk_bitmap)) {
r_ptr_int += pblk->min_write_pgs;
ppa = addr_to_gen_ppa(pblk, r_ptr_int, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
}
for (j = 0; j < pblk->min_write_pgs; j++, i++, r_ptr_int++)
@@ -411,12 +411,12 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
int pos;
w_ptr = pblk_alloc_page(pblk, line, pblk->min_write_pgs);
- ppa = addr_to_pblk_ppa(pblk, w_ptr, line->id);
+ ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
pos = pblk_ppa_to_pos(geo, ppa);
while (test_bit(pos, line->blk_bitmap)) {
w_ptr += pblk->min_write_pgs;
- ppa = addr_to_pblk_ppa(pblk, w_ptr, line->id);
+ ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
pos = pblk_ppa_to_pos(geo, ppa);
}
@@ -541,12 +541,12 @@ static int pblk_recov_scan_all_oob(struct pblk *pblk, struct pblk_line *line,
w_ptr = pblk_alloc_page(pblk, line, pblk->min_write_pgs);
ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
while (test_bit(pos, line->blk_bitmap)) {
w_ptr += pblk->min_write_pgs;
ppa = addr_to_gen_ppa(pblk, w_ptr, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
}
for (j = 0; j < pblk->min_write_pgs; j++, i++, w_ptr++)
@@ -672,12 +672,12 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line,
paddr = pblk_alloc_page(pblk, line, pblk->min_write_pgs);
ppa = addr_to_gen_ppa(pblk, paddr, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
while (test_bit(pos, line->blk_bitmap)) {
paddr += pblk->min_write_pgs;
ppa = addr_to_gen_ppa(pblk, paddr, line->id);
- pos = pblk_dev_ppa_to_pos(geo, ppa);
+ pos = pblk_ppa_to_pos(geo, ppa);
}
for (j = 0; j < pblk->min_write_pgs; j++, i++, paddr++)
@@ -817,7 +817,7 @@ static u64 pblk_line_emeta_start(struct pblk *pblk, struct pblk_line *line)
while (emeta_secs) {
emeta_start--;
- ppa = addr_to_pblk_ppa(pblk, emeta_start, line->id);
+ ppa = addr_to_gen_ppa(pblk, emeta_start, line->id);
pos = pblk_ppa_to_pos(geo, ppa);
if (!test_bit(pos, line->blk_bitmap))
emeta_secs--;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index c150728..d68a94d 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -910,12 +910,7 @@ static inline int pblk_pad_distance(struct pblk *pblk)
return NVM_MEM_PAGE_WRITE * geo->all_luns * geo->sec_per_pl;
}
-static inline int pblk_dev_ppa_to_line(struct ppa_addr p)
-{
- return p.g.blk;
-}
-
-static inline int pblk_tgt_ppa_to_line(struct ppa_addr p)
+static inline int pblk_ppa_to_line(struct ppa_addr p)
{
return p.g.blk;
}
@@ -925,10 +920,34 @@ static inline int pblk_ppa_to_pos(struct nvm_geo *geo, struct ppa_addr p)
return p.g.lun * geo->nr_chnls + p.g.ch;
}
-/* A block within a line corresponds to the lun */
-static inline int pblk_dev_ppa_to_pos(struct nvm_geo *geo, struct ppa_addr p)
+static inline struct ppa_addr addr_to_gen_ppa(struct pblk *pblk, u64 paddr,
+ u64 line_id)
{
- return p.g.lun * geo->nr_chnls + p.g.ch;
+ struct ppa_addr ppa;
+
+ ppa.ppa = 0;
+ ppa.g.blk = line_id;
+ ppa.g.pg = (paddr & pblk->ppaf.pg_mask) >> pblk->ppaf.pg_offset;
+ ppa.g.lun = (paddr & pblk->ppaf.lun_mask) >> pblk->ppaf.lun_offset;
+ ppa.g.ch = (paddr & pblk->ppaf.ch_mask) >> pblk->ppaf.ch_offset;
+ ppa.g.pl = (paddr & pblk->ppaf.pln_mask) >> pblk->ppaf.pln_offset;
+ ppa.g.sec = (paddr & pblk->ppaf.sec_mask) >> pblk->ppaf.sec_offset;
+
+ return ppa;
+}
+
+static inline u64 pblk_dev_ppa_to_line_addr(struct pblk *pblk,
+ struct ppa_addr p)
+{
+ u64 paddr;
+
+ paddr = (u64)p.g.pg << pblk->ppaf.pg_offset;
+ paddr |= (u64)p.g.lun << pblk->ppaf.lun_offset;
+ paddr |= (u64)p.g.ch << pblk->ppaf.ch_offset;
+ paddr |= (u64)p.g.pl << pblk->ppaf.pln_offset;
+ paddr |= (u64)p.g.sec << pblk->ppaf.sec_offset;
+
+ return paddr;
}
static inline struct ppa_addr pblk_ppa32_to_ppa64(struct pblk *pblk, u32 ppa32)
@@ -960,24 +979,6 @@ static inline struct ppa_addr pblk_ppa32_to_ppa64(struct pblk *pblk, u32 ppa32)
return ppa64;
}
-static inline struct ppa_addr pblk_trans_map_get(struct pblk *pblk,
- sector_t lba)
-{
- struct ppa_addr ppa;
-
- if (pblk->ppaf_bitsize < 32) {
- u32 *map = (u32 *)pblk->trans_map;
-
- ppa = pblk_ppa32_to_ppa64(pblk, map[lba]);
- } else {
- struct ppa_addr *map = (struct ppa_addr *)pblk->trans_map;
-
- ppa = map[lba];
- }
-
- return ppa;
-}
-
static inline u32 pblk_ppa64_to_ppa32(struct pblk *pblk, struct ppa_addr ppa64)
{
u32 ppa32 = 0;
@@ -999,6 +1000,24 @@ static inline u32 pblk_ppa64_to_ppa32(struct pblk *pblk, struct ppa_addr ppa64)
return ppa32;
}
+static inline struct ppa_addr pblk_trans_map_get(struct pblk *pblk,
+ sector_t lba)
+{
+ struct ppa_addr ppa;
+
+ if (pblk->ppaf_bitsize < 32) {
+ u32 *map = (u32 *)pblk->trans_map;
+
+ ppa = pblk_ppa32_to_ppa64(pblk, map[lba]);
+ } else {
+ struct ppa_addr *map = (struct ppa_addr *)pblk->trans_map;
+
+ ppa = map[lba];
+ }
+
+ return ppa;
+}
+
static inline void pblk_trans_map_set(struct pblk *pblk, sector_t lba,
struct ppa_addr ppa)
{
@@ -1013,21 +1032,6 @@ static inline void pblk_trans_map_set(struct pblk *pblk, sector_t lba,
}
}
-static inline u64 pblk_dev_ppa_to_line_addr(struct pblk *pblk,
- struct ppa_addr p)
-{
- u64 paddr;
-
- paddr = 0;
- paddr |= (u64)p.g.pg << pblk->ppaf.pg_offset;
- paddr |= (u64)p.g.lun << pblk->ppaf.lun_offset;
- paddr |= (u64)p.g.ch << pblk->ppaf.ch_offset;
- paddr |= (u64)p.g.pl << pblk->ppaf.pln_offset;
- paddr |= (u64)p.g.sec << pblk->ppaf.sec_offset;
-
- return paddr;
-}
-
static inline int pblk_ppa_empty(struct ppa_addr ppa_addr)
{
return (ppa_addr.ppa == ADDR_EMPTY);
@@ -1066,32 +1070,6 @@ static inline struct ppa_addr pblk_cacheline_to_addr(int addr)
return p;
}
-static inline struct ppa_addr addr_to_gen_ppa(struct pblk *pblk, u64 paddr,
- u64 line_id)
-{
- struct ppa_addr ppa;
-
- ppa.ppa = 0;
- ppa.g.blk = line_id;
- ppa.g.pg = (paddr & pblk->ppaf.pg_mask) >> pblk->ppaf.pg_offset;
- ppa.g.lun = (paddr & pblk->ppaf.lun_mask) >> pblk->ppaf.lun_offset;
- ppa.g.ch = (paddr & pblk->ppaf.ch_mask) >> pblk->ppaf.ch_offset;
- ppa.g.pl = (paddr & pblk->ppaf.pln_mask) >> pblk->ppaf.pln_offset;
- ppa.g.sec = (paddr & pblk->ppaf.sec_mask) >> pblk->ppaf.sec_offset;
-
- return ppa;
-}
-
-static inline struct ppa_addr addr_to_pblk_ppa(struct pblk *pblk, u64 paddr,
- u64 line_id)
-{
- struct ppa_addr ppa;
-
- ppa = addr_to_gen_ppa(pblk, paddr, line_id);
-
- return ppa;
-}
-
static inline u32 pblk_calc_meta_header_crc(struct pblk *pblk,
struct line_header *header)
{
@@ -1245,7 +1223,7 @@ static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
for (i = 0; i < rqd->nr_ppas; i++) {
ppa = ppa_list[i];
- line = &pblk->lines[pblk_dev_ppa_to_line(ppa)];
+ line = &pblk->lines[pblk_ppa_to_line(ppa)];
spin_lock(&line->lock);
if (line->state != PBLK_LINESTATE_OPEN) {
@@ -1288,11 +1266,6 @@ static inline unsigned int pblk_get_secs(struct bio *bio)
return bio->bi_iter.bi_size / PBLK_EXPOSED_PAGE_SIZE;
}
-static inline sector_t pblk_get_sector(sector_t lba)
-{
- return lba * NR_PHY_IN_LOG;
-}
-
static inline void pblk_setup_uuid(struct pblk *pblk)
{
uuid_le uuid;
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 11/25] lightnvm: pblk: remove pblk_for_each_lun helper
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (9 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 10/25] lightnvm: pblk: compress and reorder helper functions Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 12/25] lightnvm: pblk: refactor emeta consistency check Matias Bjørling
` (14 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk.h | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index d68a94d..1dbb0bf 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -51,10 +51,6 @@
#define NR_PHY_IN_LOG (PBLK_EXPOSED_PAGE_SIZE / PBLK_SECTOR)
-#define pblk_for_each_lun(pblk, rlun, i) \
- for ((i) = 0, rlun = &(pblk)->luns[0]; \
- (i) < (pblk)->nr_luns; (i)++, rlun = &(pblk)->luns[(i)])
-
/* Static pool sizes */
#define PBLK_GEN_WS_POOL_SIZE (2)
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 12/25] lightnvm: pblk: refactor emeta consistency check
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (10 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 11/25] lightnvm: pblk: remove pblk_for_each_lun helper Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 13/25] lightnvm: pblk: rename sync_point to flush_point Matias Bjørling
` (13 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Hans Holmberg, Javier González,
Matias Bjørling
From: Hans Holmberg <hans.holmberg@cnexlabs.com>
Currently pblk_recov_get_lba list does two separate things:
it checks the consistency of the emeta and extracts the lba list.
This patch separates the consistency check to make the code easier
to read and to prepare for version checks of the line emeta
persistent data format version.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-gc.c | 9 ++++++++-
drivers/lightnvm/pblk-recovery.c | 15 ++++++++++-----
drivers/lightnvm/pblk.h | 2 +-
3 files changed, 19 insertions(+), 7 deletions(-)
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index 6f0ff2c..a270337 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -169,7 +169,14 @@ static void pblk_gc_line_prepare_ws(struct work_struct *work)
* the line untouched. TODO: Implement a recovery routine that scans and
* moves all sectors on the line.
*/
- lba_list = pblk_recov_get_lba_list(pblk, emeta_buf);
+
+ ret = pblk_recov_check_emeta(pblk, emeta_buf);
+ if (ret) {
+ pr_err("pblk: inconsistent emeta (line %d)\n", line->id);
+ goto fail_free_emeta;
+ }
+
+ lba_list = emeta_to_lbas(pblk, emeta_buf);
if (!lba_list) {
pr_err("pblk: could not interpret emeta (line %d)\n", line->id);
goto fail_free_emeta;
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 1b272ae..39a2e19 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -111,18 +111,18 @@ int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
return 0;
}
-__le64 *pblk_recov_get_lba_list(struct pblk *pblk, struct line_emeta *emeta_buf)
+int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta_buf)
{
u32 crc;
crc = pblk_calc_emeta_crc(pblk, emeta_buf);
if (le32_to_cpu(emeta_buf->crc) != crc)
- return NULL;
+ return 1;
if (le32_to_cpu(emeta_buf->header.identifier) != PBLK_MAGIC)
- return NULL;
+ return 1;
- return emeta_to_lbas(pblk, emeta_buf);
+ return 0;
}
static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
@@ -137,7 +137,7 @@ static int pblk_recov_l2p_from_emeta(struct pblk *pblk, struct pblk_line *line)
u64 nr_valid_lbas, nr_lbas = 0;
u64 i;
- lba_list = pblk_recov_get_lba_list(pblk, emeta_buf);
+ lba_list = emeta_to_lbas(pblk, emeta_buf);
if (!lba_list)
return 1;
@@ -938,6 +938,11 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
goto next;
}
+ if (pblk_recov_check_emeta(pblk, line->emeta->buf)) {
+ pblk_recov_l2p_from_oob(pblk, line);
+ goto next;
+ }
+
if (pblk_recov_l2p_from_emeta(pblk, line))
pblk_recov_l2p_from_oob(pblk, line);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 1dbb0bf..8851b18 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -808,7 +808,7 @@ int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
void pblk_submit_rec(struct work_struct *work);
struct pblk_line *pblk_recov_l2p(struct pblk *pblk);
int pblk_recov_pad(struct pblk *pblk);
-__le64 *pblk_recov_get_lba_list(struct pblk *pblk, struct line_emeta *emeta);
+int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta);
int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
struct pblk_rec_ctx *recovery, u64 *comp_bits,
unsigned int comp);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 13/25] lightnvm: pblk: rename sync_point to flush_point
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (11 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 12/25] lightnvm: pblk: refactor emeta consistency check Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 14/25] lightnvm: pblk: clear flush point on completed writes Matias Bjørling
` (12 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Hans Holmberg, Javier González,
Matias Bjørling
From: Hans Holmberg <hans.holmberg@cnexlabs.com>
Sync point is a really confusing name for keeping track of
the last entry that needs to be flushed so change the name
to to flush_point instead.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-rb.c | 61 ++++++++++++++++++++++---------------------
drivers/lightnvm/pblk-write.c | 2 +-
drivers/lightnvm/pblk.h | 6 ++---
3 files changed, 35 insertions(+), 34 deletions(-)
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 62db408..941842e 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -54,7 +54,7 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
rb->seg_size = (1 << power_seg_sz);
rb->nr_entries = (1 << power_size);
rb->mem = rb->subm = rb->sync = rb->l2p_update = 0;
- rb->sync_point = EMPTY_ENTRY;
+ rb->flush_point = EMPTY_ENTRY;
spin_lock_init(&rb->w_lock);
spin_lock_init(&rb->s_lock);
@@ -112,7 +112,7 @@ int pblk_rb_init(struct pblk_rb *rb, struct pblk_rb_entry *rb_entry_base,
up_write(&pblk_rb_lock);
#ifdef CONFIG_NVM_DEBUG
- atomic_set(&rb->inflight_sync_point, 0);
+ atomic_set(&rb->inflight_flush_point, 0);
#endif
/*
@@ -349,26 +349,26 @@ void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
smp_store_release(&entry->w_ctx.flags, flags);
}
-static int pblk_rb_sync_point_set(struct pblk_rb *rb, struct bio *bio,
+static int pblk_rb_flush_point_set(struct pblk_rb *rb, struct bio *bio,
unsigned int pos)
{
struct pblk_rb_entry *entry;
- unsigned int subm, sync_point;
+ unsigned int subm, flush_point;
subm = READ_ONCE(rb->subm);
#ifdef CONFIG_NVM_DEBUG
- atomic_inc(&rb->inflight_sync_point);
+ atomic_inc(&rb->inflight_flush_point);
#endif
if (pos == subm)
return 0;
- sync_point = (pos == 0) ? (rb->nr_entries - 1) : (pos - 1);
- entry = &rb->entries[sync_point];
+ flush_point = (pos == 0) ? (rb->nr_entries - 1) : (pos - 1);
+ entry = &rb->entries[flush_point];
- /* Protect syncs */
- smp_store_release(&rb->sync_point, sync_point);
+ /* Protect flush points */
+ smp_store_release(&rb->flush_point, flush_point);
if (!bio)
return 0;
@@ -416,7 +416,7 @@ void pblk_rb_flush(struct pblk_rb *rb)
struct pblk *pblk = container_of(rb, struct pblk, rwb);
unsigned int mem = READ_ONCE(rb->mem);
- if (pblk_rb_sync_point_set(rb, NULL, mem))
+ if (pblk_rb_flush_point_set(rb, NULL, mem))
return;
pblk_write_should_kick(pblk);
@@ -440,7 +440,7 @@ static int pblk_rb_may_write_flush(struct pblk_rb *rb, unsigned int nr_entries,
#ifdef CONFIG_NVM_DEBUG
atomic_long_inc(&pblk->nr_flush);
#endif
- if (pblk_rb_sync_point_set(&pblk->rwb, bio, mem))
+ if (pblk_rb_flush_point_set(&pblk->rwb, bio, mem))
*io_ret = NVM_IO_OK;
}
@@ -607,17 +607,18 @@ unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
}
if (flags & PBLK_FLUSH_ENTRY) {
- unsigned int sync_point;
+ unsigned int flush_point;
- sync_point = READ_ONCE(rb->sync_point);
- if (sync_point == pos) {
- /* Protect syncs */
- smp_store_release(&rb->sync_point, EMPTY_ENTRY);
+ flush_point = READ_ONCE(rb->flush_point);
+ if (flush_point == pos) {
+ /* Protect flush points */
+ smp_store_release(&rb->flush_point,
+ EMPTY_ENTRY);
}
flags &= ~PBLK_FLUSH_ENTRY;
#ifdef CONFIG_NVM_DEBUG
- atomic_dec(&rb->inflight_sync_point);
+ atomic_dec(&rb->inflight_flush_point);
#endif
}
@@ -746,20 +747,20 @@ unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries)
return sync;
}
-unsigned int pblk_rb_sync_point_count(struct pblk_rb *rb)
+unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb)
{
- unsigned int subm, sync_point;
+ unsigned int subm, flush_point;
unsigned int count;
- /* Protect syncs */
- sync_point = smp_load_acquire(&rb->sync_point);
- if (sync_point == EMPTY_ENTRY)
+ /* Protect flush points */
+ flush_point = smp_load_acquire(&rb->flush_point);
+ if (flush_point == EMPTY_ENTRY)
return 0;
subm = READ_ONCE(rb->subm);
/* The sync point itself counts as a sector to sync */
- count = pblk_rb_ring_count(sync_point, subm, rb->nr_entries) + 1;
+ count = pblk_rb_ring_count(flush_point, subm, rb->nr_entries) + 1;
return count;
}
@@ -801,7 +802,7 @@ int pblk_rb_tear_down_check(struct pblk_rb *rb)
if ((rb->mem == rb->subm) && (rb->subm == rb->sync) &&
(rb->sync == rb->l2p_update) &&
- (rb->sync_point == EMPTY_ENTRY)) {
+ (rb->flush_point == EMPTY_ENTRY)) {
goto out;
}
@@ -848,7 +849,7 @@ ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf)
queued_entries++;
spin_unlock_irq(&rb->s_lock);
- if (rb->sync_point != EMPTY_ENTRY)
+ if (rb->flush_point != EMPTY_ENTRY)
offset = scnprintf(buf, PAGE_SIZE,
"%u\t%u\t%u\t%u\t%u\t%u\t%u - %u/%u/%u - %d\n",
rb->nr_entries,
@@ -857,14 +858,14 @@ ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf)
rb->sync,
rb->l2p_update,
#ifdef CONFIG_NVM_DEBUG
- atomic_read(&rb->inflight_sync_point),
+ atomic_read(&rb->inflight_flush_point),
#else
0,
#endif
- rb->sync_point,
+ rb->flush_point,
pblk_rb_read_count(rb),
pblk_rb_space(rb),
- pblk_rb_sync_point_count(rb),
+ pblk_rb_flush_point_count(rb),
queued_entries);
else
offset = scnprintf(buf, PAGE_SIZE,
@@ -875,13 +876,13 @@ ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf)
rb->sync,
rb->l2p_update,
#ifdef CONFIG_NVM_DEBUG
- atomic_read(&rb->inflight_sync_point),
+ atomic_read(&rb->inflight_flush_point),
#else
0,
#endif
pblk_rb_read_count(rb),
pblk_rb_space(rb),
- pblk_rb_sync_point_count(rb),
+ pblk_rb_flush_point_count(rb),
queued_entries);
return offset;
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 6c30b7a..018af87 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -508,7 +508,7 @@ static int pblk_submit_write(struct pblk *pblk)
if (!secs_avail)
return 1;
- secs_to_flush = pblk_rb_sync_point_count(&pblk->rwb);
+ secs_to_flush = pblk_rb_flush_point_count(&pblk->rwb);
if (!secs_to_flush && secs_avail < pblk->min_write_pgs)
return 1;
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 8851b18..5f7cd6f 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -166,7 +166,7 @@ struct pblk_rb {
* the last submitted entry that has
* been successfully persisted to media
*/
- unsigned int sync_point; /* Sync point - last entry that must be
+ unsigned int flush_point; /* Sync point - last entry that must be
* flushed to the media. Used with
* REQ_FLUSH and REQ_FUA
*/
@@ -189,7 +189,7 @@ struct pblk_rb {
spinlock_t s_lock; /* Sync lock */
#ifdef CONFIG_NVM_DEBUG
- atomic_t inflight_sync_point; /* Not served REQ_FLUSH | REQ_FUA */
+ atomic_t inflight_flush_point; /* Not served REQ_FLUSH | REQ_FUA */
#endif
};
@@ -687,7 +687,7 @@ unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries);
struct pblk_rb_entry *pblk_rb_sync_scan_entry(struct pblk_rb *rb,
struct ppa_addr *ppa);
void pblk_rb_sync_end(struct pblk_rb *rb, unsigned long *flags);
-unsigned int pblk_rb_sync_point_count(struct pblk_rb *rb);
+unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb);
unsigned int pblk_rb_read_count(struct pblk_rb *rb);
unsigned int pblk_rb_sync_count(struct pblk_rb *rb);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 14/25] lightnvm: pblk: clear flush point on completed writes
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (12 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 13/25] lightnvm: pblk: rename sync_point to flush_point Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 15/25] lightnvm: pblk: prevent premature sync point resets Matias Bjørling
` (11 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Hans Holmberg, Hans Holmberg,
Javier González, Matias Bjørling
From: Hans Holmberg <hans@owltronix.com>
Move completion of syncs and clearing of flush points to the
write completion path - this ensures that the data has been
comitted to the media before completing bios containing syncs.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-rb.c | 58 +++++++++++++++++++++----------------------
drivers/lightnvm/pblk-write.c | 17 ++++++++++++-
2 files changed, 44 insertions(+), 31 deletions(-)
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 941842e..672ef8c 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -353,17 +353,17 @@ static int pblk_rb_flush_point_set(struct pblk_rb *rb, struct bio *bio,
unsigned int pos)
{
struct pblk_rb_entry *entry;
- unsigned int subm, flush_point;
+ unsigned int sync, flush_point;
- subm = READ_ONCE(rb->subm);
+ sync = READ_ONCE(rb->sync);
+
+ if (pos == sync)
+ return 0;
#ifdef CONFIG_NVM_DEBUG
atomic_inc(&rb->inflight_flush_point);
#endif
- if (pos == subm)
- return 0;
-
flush_point = (pos == 0) ? (rb->nr_entries - 1) : (pos - 1);
entry = &rb->entries[flush_point];
@@ -606,22 +606,6 @@ unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
return NVM_IO_ERR;
}
- if (flags & PBLK_FLUSH_ENTRY) {
- unsigned int flush_point;
-
- flush_point = READ_ONCE(rb->flush_point);
- if (flush_point == pos) {
- /* Protect flush points */
- smp_store_release(&rb->flush_point,
- EMPTY_ENTRY);
- }
-
- flags &= ~PBLK_FLUSH_ENTRY;
-#ifdef CONFIG_NVM_DEBUG
- atomic_dec(&rb->inflight_flush_point);
-#endif
- }
-
flags &= ~PBLK_WRITTEN_DATA;
flags |= PBLK_SUBMITTED_ENTRY;
@@ -731,15 +715,24 @@ void pblk_rb_sync_end(struct pblk_rb *rb, unsigned long *flags)
unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries)
{
- unsigned int sync;
- unsigned int i;
-
+ unsigned int sync, flush_point;
lockdep_assert_held(&rb->s_lock);
sync = READ_ONCE(rb->sync);
+ flush_point = READ_ONCE(rb->flush_point);
- for (i = 0; i < nr_entries; i++)
- sync = (sync + 1) & (rb->nr_entries - 1);
+ if (flush_point != EMPTY_ENTRY) {
+ unsigned int secs_to_flush;
+
+ secs_to_flush = pblk_rb_ring_count(flush_point, sync,
+ rb->nr_entries);
+ if (secs_to_flush < nr_entries) {
+ /* Protect flush points */
+ smp_store_release(&rb->flush_point, EMPTY_ENTRY);
+ }
+ }
+
+ sync = (sync + nr_entries) & (rb->nr_entries - 1);
/* Protect from counts */
smp_store_release(&rb->sync, sync);
@@ -747,22 +740,27 @@ unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries)
return sync;
}
+/* Calculate how many sectors to submit up to the current flush point. */
unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb)
{
- unsigned int subm, flush_point;
- unsigned int count;
+ unsigned int subm, sync, flush_point;
+ unsigned int submitted, to_flush;
/* Protect flush points */
flush_point = smp_load_acquire(&rb->flush_point);
if (flush_point == EMPTY_ENTRY)
return 0;
+ /* Protect syncs */
+ sync = smp_load_acquire(&rb->sync);
+
subm = READ_ONCE(rb->subm);
+ submitted = pblk_rb_ring_count(subm, sync, rb->nr_entries);
/* The sync point itself counts as a sector to sync */
- count = pblk_rb_ring_count(flush_point, subm, rb->nr_entries) + 1;
+ to_flush = pblk_rb_ring_count(flush_point, sync, rb->nr_entries) + 1;
- return count;
+ return (submitted < to_flush) ? (to_flush - submitted) : 0;
}
/*
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 018af87..aae86ed 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -21,13 +21,28 @@ static unsigned long pblk_end_w_bio(struct pblk *pblk, struct nvm_rq *rqd,
struct pblk_c_ctx *c_ctx)
{
struct bio *original_bio;
+ struct pblk_rb *rwb = &pblk->rwb;
unsigned long ret;
int i;
for (i = 0; i < c_ctx->nr_valid; i++) {
struct pblk_w_ctx *w_ctx;
+ int pos = c_ctx->sentry + i;
+ int flags;
+
+ w_ctx = pblk_rb_w_ctx(rwb, pos);
+ flags = READ_ONCE(w_ctx->flags);
+
+ if (flags & PBLK_FLUSH_ENTRY) {
+ flags &= ~PBLK_FLUSH_ENTRY;
+ /* Release flags on context. Protect from writes */
+ smp_store_release(&w_ctx->flags, flags);
+
+#ifdef CONFIG_NVM_DEBUG
+ atomic_dec(&rwb->inflight_flush_point);
+#endif
+ }
- w_ctx = pblk_rb_w_ctx(&pblk->rwb, c_ctx->sentry + i);
while ((original_bio = bio_list_pop(&w_ctx->bios)))
bio_endio(original_bio);
}
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 15/25] lightnvm: pblk: prevent premature sync point resets
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (13 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 14/25] lightnvm: pblk: clear flush point on completed writes Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 16/25] lightnvm: pblk: remove pblk_gc_stop Matias Bjørling
` (10 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Hans Holmberg, Javier González,
Matias Bjørling
From: Hans Holmberg <hans.holmberg@cnexlabs.com>
Unless we protect flush pointer updates with a lock, we risk
resetting new flush points before we've synced all sectors
up to that point.
This patch protects new flush points with the same spin lock
that is being held when advancing the sync pointer and
resetting completed flush points.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-rb.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/lightnvm/pblk-rb.c b/drivers/lightnvm/pblk-rb.c
index 672ef8c..ec8fc31 100644
--- a/drivers/lightnvm/pblk-rb.c
+++ b/drivers/lightnvm/pblk-rb.c
@@ -367,17 +367,17 @@ static int pblk_rb_flush_point_set(struct pblk_rb *rb, struct bio *bio,
flush_point = (pos == 0) ? (rb->nr_entries - 1) : (pos - 1);
entry = &rb->entries[flush_point];
+ pblk_rb_sync_init(rb, NULL);
+
/* Protect flush points */
smp_store_release(&rb->flush_point, flush_point);
- if (!bio)
- return 0;
+ if (bio)
+ bio_list_add(&entry->w_ctx.bios, bio);
- spin_lock_irq(&rb->s_lock);
- bio_list_add(&entry->w_ctx.bios, bio);
- spin_unlock_irq(&rb->s_lock);
+ pblk_rb_sync_end(rb, NULL);
- return 1;
+ return bio ? 1 : 0;
}
static int __pblk_rb_may_write(struct pblk_rb *rb, unsigned int nr_entries,
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 16/25] lightnvm: pblk: remove pblk_gc_stop
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (14 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 15/25] lightnvm: pblk: prevent premature sync point resets Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 17/25] lightnvm: pblk: use exact free block counter in RL Matias Bjørling
` (9 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Hans Holmberg, Javier González,
Matias Bjørling
From: Hans Holmberg <hans.holmberg@cnexlabs.com>
pblk_gc_stop just sets pblk->gc->gc_active to zero, ignoring
the flush parameter. This is plain confusing, so remove the
function and set the gc active flag at the call points instead.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-gc.c | 14 ++------------
1 file changed, 2 insertions(+), 12 deletions(-)
diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index a270337..278bc02 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -526,22 +526,12 @@ void pblk_gc_should_start(struct pblk *pblk)
}
}
-/*
- * If flush_wq == 1 then no lock should be held by the caller since
- * flush_workqueue can sleep
- */
-static void pblk_gc_stop(struct pblk *pblk, int flush_wq)
-{
- pblk->gc.gc_active = 0;
- pr_debug("pblk: gc stop\n");
-}
-
void pblk_gc_should_stop(struct pblk *pblk)
{
struct pblk_gc *gc = &pblk->gc;
if (gc->gc_active && !gc->gc_forced)
- pblk_gc_stop(pblk, 0);
+ gc->gc_active = 0;
}
void pblk_gc_should_kick(struct pblk *pblk)
@@ -667,7 +657,7 @@ void pblk_gc_exit(struct pblk *pblk)
gc->gc_enabled = 0;
del_timer_sync(&gc->gc_timer);
- pblk_gc_stop(pblk, 1);
+ gc->gc_active = 0;
if (gc->gc_ts)
kthread_stop(gc->gc_ts);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 17/25] lightnvm: pblk: use exact free block counter in RL
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (15 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 16/25] lightnvm: pblk: remove pblk_gc_stop Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl Matias Bjørling
` (8 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Hans Holmberg,
Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Until now, pblk's rate-limiter has used a heuristic to reserve space for
GC I/O given that the over-provision area was fixed.
In preparation for allowing to define the over-provision area on target
creation, define a dedicated free_block counter in the rate-limiter to
track the number of blocks being used for user data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-core.c | 19 +++++---------
drivers/lightnvm/pblk-init.c | 18 +++++++++++---
drivers/lightnvm/pblk-recovery.c | 4 +--
drivers/lightnvm/pblk-rl.c | 54 +++++++++++++++++++++++++++-------------
drivers/lightnvm/pblk-sysfs.c | 9 ++++---
drivers/lightnvm/pblk.h | 15 ++++++-----
6 files changed, 73 insertions(+), 46 deletions(-)
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 54d0cef..5ec7ad6 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -1145,7 +1145,7 @@ int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line)
}
spin_unlock(&l_mg->free_lock);
- pblk_rl_free_lines_dec(&pblk->rl, line);
+ pblk_rl_free_lines_dec(&pblk->rl, line, true);
if (!pblk_line_init_bb(pblk, line, 0)) {
list_add(&line->list, &l_mg->free_list);
@@ -1233,7 +1233,7 @@ static struct pblk_line *pblk_line_retry(struct pblk *pblk,
l_mg->data_line = retry_line;
spin_unlock(&l_mg->free_lock);
- pblk_rl_free_lines_dec(&pblk->rl, retry_line);
+ pblk_rl_free_lines_dec(&pblk->rl, line, false);
if (pblk_line_erase(pblk, retry_line))
goto retry;
@@ -1252,7 +1252,6 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
{
struct pblk_line_mgmt *l_mg = &pblk->l_mg;
struct pblk_line *line;
- int is_next = 0;
spin_lock(&l_mg->free_lock);
line = pblk_line_get(pblk);
@@ -1280,7 +1279,6 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
} else {
l_mg->data_next->seq_nr = l_mg->d_seq_nr++;
l_mg->data_next->type = PBLK_LINETYPE_DATA;
- is_next = 1;
}
spin_unlock(&l_mg->free_lock);
@@ -1290,10 +1288,6 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
return NULL;
}
- pblk_rl_free_lines_dec(&pblk->rl, line);
- if (is_next)
- pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
-
retry_setup:
if (!pblk_line_init_metadata(pblk, line, NULL)) {
line = pblk_line_retry(pblk, line);
@@ -1311,6 +1305,8 @@ struct pblk_line *pblk_line_get_first_data(struct pblk *pblk)
goto retry_setup;
}
+ pblk_rl_free_lines_dec(&pblk->rl, line, true);
+
return line;
}
@@ -1395,7 +1391,6 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
struct pblk_line_mgmt *l_mg = &pblk->l_mg;
struct pblk_line *cur, *new = NULL;
unsigned int left_seblks;
- int is_next = 0;
cur = l_mg->data_line;
new = l_mg->data_next;
@@ -1444,6 +1439,8 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
goto retry_setup;
}
+ pblk_rl_free_lines_dec(&pblk->rl, new, true);
+
/* Allocate next line for preparation */
spin_lock(&l_mg->free_lock);
l_mg->data_next = pblk_line_get(pblk);
@@ -1457,13 +1454,9 @@ struct pblk_line *pblk_line_replace_data(struct pblk *pblk)
} else {
l_mg->data_next->seq_nr = l_mg->d_seq_nr++;
l_mg->data_next->type = PBLK_LINETYPE_DATA;
- is_next = 1;
}
spin_unlock(&l_mg->free_lock);
- if (is_next)
- pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
-
out:
return new;
}
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index d13bb51..c8a7182 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -579,22 +579,34 @@ static unsigned int calc_emeta_len(struct pblk *pblk)
static void pblk_set_provision(struct pblk *pblk, long nr_free_blks)
{
struct nvm_tgt_dev *dev = pblk->dev;
+ struct pblk_line_mgmt *l_mg = &pblk->l_mg;
+ struct pblk_line_meta *lm = &pblk->lm;
struct nvm_geo *geo = &dev->geo;
sector_t provisioned;
+ int sec_meta, blk_meta;
- pblk->over_pct = 20;
+ pblk->op = 20;
provisioned = nr_free_blks;
- provisioned *= (100 - pblk->over_pct);
+ provisioned *= (100 - pblk->op);
sector_div(provisioned, 100);
+ pblk->op_blks = nr_free_blks - provisioned;
+
/* Internally pblk manages all free blocks, but all calculations based
* on user capacity consider only provisioned blocks
*/
pblk->rl.total_blocks = nr_free_blks;
pblk->rl.nr_secs = nr_free_blks * geo->sec_per_chk;
- pblk->capacity = provisioned * geo->sec_per_chk;
+
+ /* Consider sectors used for metadata */
+ sec_meta = (lm->smeta_sec + lm->emeta_sec[0]) * l_mg->nr_free_lines;
+ blk_meta = DIV_ROUND_UP(sec_meta, geo->sec_per_chk);
+
+ pblk->capacity = (provisioned - blk_meta) * geo->sec_per_chk;
+
atomic_set(&pblk->rl.free_blocks, nr_free_blks);
+ atomic_set(&pblk->rl.free_user_blocks, nr_free_blks);
}
static int pblk_lines_alloc_metadata(struct pblk *pblk)
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index 39a2e19..fd38036 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -989,10 +989,8 @@ struct pblk_line *pblk_recov_l2p(struct pblk *pblk)
}
spin_unlock(&l_mg->free_lock);
- if (is_next) {
+ if (is_next)
pblk_line_erase(pblk, l_mg->data_next);
- pblk_rl_free_lines_dec(&pblk->rl, l_mg->data_next);
- }
out:
if (found_lines != recovered_lines)
diff --git a/drivers/lightnvm/pblk-rl.c b/drivers/lightnvm/pblk-rl.c
index dacc719..0d457b1 100644
--- a/drivers/lightnvm/pblk-rl.c
+++ b/drivers/lightnvm/pblk-rl.c
@@ -89,17 +89,15 @@ unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl)
return atomic_read(&rl->free_blocks);
}
-/*
- * We check for (i) the number of free blocks in the current LUN and (ii) the
- * total number of free blocks in the pblk instance. This is to even out the
- * number of free blocks on each LUN when GC kicks in.
- *
- * Only the total number of free blocks is used to configure the rate limiter.
- */
-void pblk_rl_update_rates(struct pblk_rl *rl)
+unsigned long pblk_rl_nr_user_free_blks(struct pblk_rl *rl)
+{
+ return atomic_read(&rl->free_user_blocks);
+}
+
+static void __pblk_rl_update_rates(struct pblk_rl *rl,
+ unsigned long free_blocks)
{
struct pblk *pblk = container_of(rl, struct pblk, rl);
- unsigned long free_blocks = pblk_rl_nr_free_blks(rl);
int max = rl->rb_budget;
if (free_blocks >= rl->high) {
@@ -132,20 +130,37 @@ void pblk_rl_update_rates(struct pblk_rl *rl)
pblk_gc_should_stop(pblk);
}
+void pblk_rl_update_rates(struct pblk_rl *rl)
+{
+ __pblk_rl_update_rates(rl, pblk_rl_nr_user_free_blks(rl));
+}
+
void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line)
{
int blk_in_line = atomic_read(&line->blk_in_line);
+ int free_blocks;
atomic_add(blk_in_line, &rl->free_blocks);
- pblk_rl_update_rates(rl);
+ free_blocks = atomic_add_return(blk_in_line, &rl->free_user_blocks);
+
+ __pblk_rl_update_rates(rl, free_blocks);
}
-void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line)
+void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line,
+ bool used)
{
int blk_in_line = atomic_read(&line->blk_in_line);
+ int free_blocks;
atomic_sub(blk_in_line, &rl->free_blocks);
- pblk_rl_update_rates(rl);
+
+ if (used)
+ free_blocks = atomic_sub_return(blk_in_line,
+ &rl->free_user_blocks);
+ else
+ free_blocks = atomic_read(&rl->free_user_blocks);
+
+ __pblk_rl_update_rates(rl, free_blocks);
}
int pblk_rl_high_thrs(struct pblk_rl *rl)
@@ -174,17 +189,22 @@ void pblk_rl_free(struct pblk_rl *rl)
void pblk_rl_init(struct pblk_rl *rl, int budget)
{
struct pblk *pblk = container_of(rl, struct pblk, rl);
+ struct nvm_tgt_dev *dev = pblk->dev;
+ struct nvm_geo *geo = &dev->geo;
+ struct pblk_line_mgmt *l_mg = &pblk->l_mg;
struct pblk_line_meta *lm = &pblk->lm;
int min_blocks = lm->blk_per_line * PBLK_GC_RSV_LINE;
+ int sec_meta, blk_meta;
+
unsigned int rb_windows;
- rl->high = rl->total_blocks / PBLK_USER_HIGH_THRS;
+ /* Consider sectors used for metadata */
+ sec_meta = (lm->smeta_sec + lm->emeta_sec[0]) * l_mg->nr_free_lines;
+ blk_meta = DIV_ROUND_UP(sec_meta, geo->sec_per_chk);
+
+ rl->high = pblk->op_blks - blk_meta - lm->blk_per_line;
rl->high_pw = get_count_order(rl->high);
- rl->low = rl->total_blocks / PBLK_USER_LOW_THRS;
- if (rl->low < min_blocks)
- rl->low = min_blocks;
-
rl->rsv_blocks = min_blocks;
/* This will always be a power-of-2 */
diff --git a/drivers/lightnvm/pblk-sysfs.c b/drivers/lightnvm/pblk-sysfs.c
index 5cee2ac..620bab8 100644
--- a/drivers/lightnvm/pblk-sysfs.c
+++ b/drivers/lightnvm/pblk-sysfs.c
@@ -49,11 +49,12 @@ static ssize_t pblk_sysfs_luns_show(struct pblk *pblk, char *page)
static ssize_t pblk_sysfs_rate_limiter(struct pblk *pblk, char *page)
{
- int free_blocks, total_blocks;
+ int free_blocks, free_user_blocks, total_blocks;
int rb_user_max, rb_user_cnt;
int rb_gc_max, rb_gc_cnt, rb_budget, rb_state;
- free_blocks = atomic_read(&pblk->rl.free_blocks);
+ free_blocks = pblk_rl_nr_free_blks(&pblk->rl);
+ free_user_blocks = pblk_rl_nr_user_free_blks(&pblk->rl);
rb_user_max = pblk->rl.rb_user_max;
rb_user_cnt = atomic_read(&pblk->rl.rb_user_cnt);
rb_gc_max = pblk->rl.rb_gc_max;
@@ -64,16 +65,16 @@ static ssize_t pblk_sysfs_rate_limiter(struct pblk *pblk, char *page)
total_blocks = pblk->rl.total_blocks;
return snprintf(page, PAGE_SIZE,
- "u:%u/%u,gc:%u/%u(%u/%u)(stop:<%u,full:>%u,free:%d/%d)-%d\n",
+ "u:%u/%u,gc:%u/%u(%u)(stop:<%u,full:>%u,free:%d/%d/%d)-%d\n",
rb_user_cnt,
rb_user_max,
rb_gc_cnt,
rb_gc_max,
rb_state,
rb_budget,
- pblk->rl.low,
pblk->rl.high,
free_blocks,
+ free_user_blocks,
total_blocks,
READ_ONCE(pblk->rl.rb_user_active));
}
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 5f7cd6f..1e719d4 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -252,9 +252,6 @@ struct pblk_rl {
unsigned int high; /* Upper threshold for rate limiter (free run -
* user I/O rate limiter
*/
- unsigned int low; /* Lower threshold for rate limiter (user I/O
- * rate limiter - stall)
- */
unsigned int high_pw; /* High rounded up as a power of 2 */
#define PBLK_USER_HIGH_THRS 8 /* Begin write limit at 12% available blks */
@@ -288,7 +285,9 @@ struct pblk_rl {
unsigned long long nr_secs;
unsigned long total_blocks;
- atomic_t free_blocks;
+
+ atomic_t free_blocks; /* Total number of free blocks (+ OP) */
+ atomic_t free_user_blocks; /* Number of user free blocks (no OP) */
};
#define PBLK_LINE_EMPTY (~0U)
@@ -579,7 +578,9 @@ struct pblk {
*/
sector_t capacity; /* Device capacity when bad blocks are subtracted */
- int over_pct; /* Percentage of device used for over-provisioning */
+
+ int op; /* Percentage of device used for over-provisioning */
+ int op_blks; /* Number of blocks used for over-provisioning */
/* pblk provisioning values. Used by rate limiter */
struct pblk_rl rl;
@@ -839,6 +840,7 @@ void pblk_rl_free(struct pblk_rl *rl);
void pblk_rl_update_rates(struct pblk_rl *rl);
int pblk_rl_high_thrs(struct pblk_rl *rl);
unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl);
+unsigned long pblk_rl_nr_user_free_blks(struct pblk_rl *rl);
int pblk_rl_user_may_insert(struct pblk_rl *rl, int nr_entries);
void pblk_rl_inserted(struct pblk_rl *rl, int nr_entries);
void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries);
@@ -847,7 +849,8 @@ void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries);
void pblk_rl_out(struct pblk_rl *rl, int nr_user, int nr_gc);
int pblk_rl_max_io(struct pblk_rl *rl);
void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line);
-void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line);
+void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line,
+ bool used);
int pblk_rl_is_limit(struct pblk_rl *rl);
/*
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (16 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 17/25] lightnvm: pblk: use exact free block counter in RL Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 19:33 ` Randy Dunlap
2018-01-05 13:16 ` [GIT PULL 19/25] lightnvm: pblk: ignore high ecc errors on recovery Matias Bjørling
` (7 subsequent siblings)
25 siblings, 1 reply; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Hans Holmberg,
Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Allow to set the over-provision percentage on target creation. In case
that the value is not provided, fall back to the default value set by
the target.
In pblk, set the default OP to 11% of the total size of the device
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/core.c | 106 +++++++++++++++++++++++++++++++++---------
drivers/lightnvm/pblk-init.c | 5 +-
drivers/lightnvm/pblk.h | 2 +
include/linux/lightnvm.h | 6 +++
include/uapi/linux/lightnvm.h | 9 ++++
5 files changed, 104 insertions(+), 24 deletions(-)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index d5f231c..dcc9e62 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -140,7 +140,8 @@ static void nvm_remove_tgt_dev(struct nvm_tgt_dev *tgt_dev, int clear)
}
static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
- int lun_begin, int lun_end)
+ u16 lun_begin, u16 lun_end,
+ u16 op)
{
struct nvm_tgt_dev *tgt_dev = NULL;
struct nvm_dev_map *dev_rmap = dev->rmap;
@@ -219,6 +220,7 @@ static struct nvm_tgt_dev *nvm_create_tgt_dev(struct nvm_dev *dev,
tgt_dev->geo.nr_chnls = nr_chnls;
tgt_dev->geo.all_luns = nr_luns;
tgt_dev->geo.nr_luns = (lun_balanced) ? prev_nr_luns : -1;
+ tgt_dev->geo.op = op;
tgt_dev->total_secs = nr_luns * tgt_dev->geo.sec_per_lun;
tgt_dev->q = dev->q;
tgt_dev->map = dev_map;
@@ -266,9 +268,57 @@ static struct nvm_tgt_type *nvm_find_target_type(const char *name)
return tt;
}
+static int nvm_config_check_luns(struct nvm_geo *geo, int lun_begin,
+ int lun_end)
+{
+ if (lun_begin > lun_end || lun_end >= geo->all_luns) {
+ pr_err("nvm: lun out of bound (%u:%u > %u)\n",
+ lun_begin, lun_end, geo->all_luns - 1);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int __nvm_config_simple(struct nvm_dev *dev,
+ struct nvm_ioctl_create_simple *s)
+{
+ struct nvm_geo *geo = &dev->geo;
+
+ if (s->lun_begin == -1 && s->lun_end == -1) {
+ s->lun_begin = 0;
+ s->lun_end = geo->all_luns - 1;
+ }
+
+ return nvm_config_check_luns(geo, s->lun_begin, s->lun_end);
+}
+
+static int __nvm_config_extended(struct nvm_dev *dev,
+ struct nvm_ioctl_create_extended *e)
+{
+ struct nvm_geo *geo = &dev->geo;
+
+ if (e->lun_begin == 0xFFFF && e->lun_end == 0xFFFF) {
+ e->lun_begin = 0;
+ e->lun_end = dev->geo.all_luns - 1;
+ }
+
+ /* op not set falls into target's default */
+ if (e->op == 0xFFFF)
+ e->op = NVM_TARGET_DEFAULT_OP;
+
+ if (e->op < NVM_TARGET_MIN_OP ||
+ e->op > NVM_TARGET_MAX_OP) {
+ pr_err("nvm: invalid over provisioning value\n");
+ return -EINVAL;
+ }
+
+ return nvm_config_check_luns(geo, e->lun_begin, e->lun_end);
+}
+
static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
{
- struct nvm_ioctl_create_simple *s = &create->conf.s;
+ struct nvm_ioctl_create_extended e;
struct request_queue *tqueue;
struct gendisk *tdisk;
struct nvm_tgt_type *tt;
@@ -277,6 +327,28 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
void *targetdata;
int ret;
+ switch (create->conf.type) {
+ case NVM_CONFIG_TYPE_SIMPLE:
+ ret = __nvm_config_simple(dev, &create->conf.s);
+ if (ret)
+ return ret;
+
+ e.lun_begin = create->conf.s.lun_begin;
+ e.lun_end = create->conf.s.lun_end;
+ e.op = NVM_TARGET_DEFAULT_OP;
+ break;
+ case NVM_CONFIG_TYPE_EXTENDED:
+ ret = __nvm_config_extended(dev, &create->conf.e);
+ if (ret)
+ return ret;
+
+ e = create->conf.e;
+ break;
+ default:
+ pr_err("nvm: config type not valid\n");
+ return -EINVAL;
+ }
+
tt = nvm_find_target_type(create->tgttype);
if (!tt) {
pr_err("nvm: target type %s not found\n", create->tgttype);
@@ -289,7 +361,7 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
return -EINVAL;
}
- ret = nvm_reserve_luns(dev, s->lun_begin, s->lun_end);
+ ret = nvm_reserve_luns(dev, e.lun_begin, e.lun_end);
if (ret)
return ret;
@@ -299,7 +371,7 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
goto err_reserve;
}
- tgt_dev = nvm_create_tgt_dev(dev, s->lun_begin, s->lun_end);
+ tgt_dev = nvm_create_tgt_dev(dev, e.lun_begin, e.lun_end, e.op);
if (!tgt_dev) {
pr_err("nvm: could not create target device\n");
ret = -ENOMEM;
@@ -369,7 +441,7 @@ static int nvm_create_tgt(struct nvm_dev *dev, struct nvm_ioctl_create *create)
err_t:
kfree(t);
err_reserve:
- nvm_release_luns_err(dev, s->lun_begin, s->lun_end);
+ nvm_release_luns_err(dev, e.lun_begin, e.lun_end);
return ret;
}
@@ -949,7 +1021,6 @@ EXPORT_SYMBOL(nvm_unregister);
static int __nvm_configure_create(struct nvm_ioctl_create *create)
{
struct nvm_dev *dev;
- struct nvm_ioctl_create_simple *s;
down_write(&nvm_lock);
dev = nvm_find_nvm_dev(create->dev);
@@ -960,23 +1031,6 @@ static int __nvm_configure_create(struct nvm_ioctl_create *create)
return -EINVAL;
}
- if (create->conf.type != NVM_CONFIG_TYPE_SIMPLE) {
- pr_err("nvm: config type not valid\n");
- return -EINVAL;
- }
- s = &create->conf.s;
-
- if (s->lun_begin == -1 && s->lun_end == -1) {
- s->lun_begin = 0;
- s->lun_end = dev->geo.all_luns - 1;
- }
-
- if (s->lun_begin > s->lun_end || s->lun_end >= dev->geo.all_luns) {
- pr_err("nvm: lun out of bound (%u:%u > %u)\n",
- s->lun_begin, s->lun_end, dev->geo.all_luns - 1);
- return -EINVAL;
- }
-
return nvm_create_tgt(dev, create);
}
@@ -1076,6 +1130,12 @@ static long nvm_ioctl_dev_create(struct file *file, void __user *arg)
if (copy_from_user(&create, arg, sizeof(struct nvm_ioctl_create)))
return -EFAULT;
+ if (create.conf.type == NVM_CONFIG_TYPE_EXTENDED &&
+ create.conf.e.rsv != 0) {
+ pr_err("nvm: reserved config field in use\n");
+ return -EINVAL;
+ }
+
create.dev[DISK_NAME_LEN - 1] = '\0';
create.tgttype[NVM_TTYPE_NAME_MAX - 1] = '\0';
create.tgtname[DISK_NAME_LEN - 1] = '\0';
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index c8a7182..533f690 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -585,7 +585,10 @@ static void pblk_set_provision(struct pblk *pblk, long nr_free_blks)
sector_t provisioned;
int sec_meta, blk_meta;
- pblk->op = 20;
+ if (geo->op == NVM_TARGET_DEFAULT_OP)
+ pblk->op = PBLK_DEFAULT_OP;
+ else
+ pblk->op = geo->op;
provisioned = nr_free_blks;
provisioned *= (100 - pblk->op);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 1e719d4..19e622c 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -54,6 +54,8 @@
/* Static pool sizes */
#define PBLK_GEN_WS_POOL_SIZE (2)
+#define PBLK_DEFAULT_OP (11)
+
enum {
PBLK_READ = READ,
PBLK_WRITE = WRITE,/* Write from write buffer */
diff --git a/include/linux/lightnvm.h b/include/linux/lightnvm.h
index 8e43bfe..7f4b60a 100644
--- a/include/linux/lightnvm.h
+++ b/include/linux/lightnvm.h
@@ -218,6 +218,10 @@ struct nvm_target {
#define ADDR_EMPTY (~0ULL)
+#define NVM_TARGET_DEFAULT_OP (101)
+#define NVM_TARGET_MIN_OP (3)
+#define NVM_TARGET_MAX_OP (80)
+
#define NVM_VERSION_MAJOR 1
#define NVM_VERSION_MINOR 0
#define NVM_VERSION_PATCH 0
@@ -291,6 +295,8 @@ struct nvm_geo {
int max_rq_size;
+ int op;
+
struct nvm_addr_format ppaf;
/* Legacy 1.2 specific geometry */
diff --git a/include/uapi/linux/lightnvm.h b/include/uapi/linux/lightnvm.h
index 42d1a43..f9a1be7 100644
--- a/include/uapi/linux/lightnvm.h
+++ b/include/uapi/linux/lightnvm.h
@@ -75,14 +75,23 @@ struct nvm_ioctl_create_simple {
__u32 lun_end;
};
+struct nvm_ioctl_create_extended {
+ __u16 lun_begin;
+ __u16 lun_end;
+ __u16 op;
+ __u16 rsv;
+};
+
enum {
NVM_CONFIG_TYPE_SIMPLE = 0,
+ NVM_CONFIG_TYPE_EXTENDED = 1,
};
struct nvm_ioctl_create_conf {
__u32 type;
union {
struct nvm_ioctl_create_simple s;
+ struct nvm_ioctl_create_extended e;
};
};
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* Re: [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 13:16 ` [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl Matias Bjørling
@ 2018-01-05 19:33 ` Randy Dunlap
2018-01-05 19:52 ` Javier Gonzalez
0 siblings, 1 reply; 45+ messages in thread
From: Randy Dunlap @ 2018-01-05 19:33 UTC (permalink / raw)
To: Matias Bjørling, axboe
Cc: linux-block, linux-kernel, Javier González, Hans Holmberg
On 01/05/2018 05:16 AM, Matias Bjørling wrote:
> From: Javier González <javier@cnexlabs.com>
>
> Allow to set the over-provision percentage on target creation. In case
> that the value is not provided, fall back to the default value set by
> the target.
>
> In pblk, set the default OP to 11% of the total size of the device
> +#define PBLK_DEFAULT_OP (11)
Hi,
Just curious -- where does 11 come from? Is it a spec value?
thanks,
--
~Randy
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 19:33 ` Randy Dunlap
@ 2018-01-05 19:52 ` Javier Gonzalez
2018-01-05 19:53 ` Matias Bjørling
0 siblings, 1 reply; 45+ messages in thread
From: Javier Gonzalez @ 2018-01-05 19:52 UTC (permalink / raw)
To: Randy Dunlap
Cc: Matias Bjørling, Jens Axboe, linux-block, linux-kernel,
Hans Holmberg (C)
[-- Attachment #1: Type: text/plain, Size: 823 bytes --]
> On 5 Jan 2018, at 20.33, Randy Dunlap <rdunlap@infradead.org> wrote:
>
> On 01/05/2018 05:16 AM, Matias Bjørling wrote:
>> From: Javier González <javier@cnexlabs.com>
>>
>> Allow to set the over-provision percentage on target creation. In case
>> that the value is not provided, fall back to the default value set by
>> the target.
>>
>> In pblk, set the default OP to 11% of the total size of the device
>
>> +#define PBLK_DEFAULT_OP (11)
>
> Hi,
> Just curious -- where does 11 come from? Is it a spec value?
11 stands for 11% over-provisioning for the media to allow garbage
collection. Different SSDs have different values based on the targeted
workload - 11% is a common default value for standard drives.
The spec. does not specify any default values of over-provisioning.
Javier.
[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 19:52 ` Javier Gonzalez
@ 2018-01-05 19:53 ` Matias Bjørling
2018-01-05 19:56 ` Javier Gonzalez
0 siblings, 1 reply; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 19:53 UTC (permalink / raw)
To: Javier Gonzalez, Randy Dunlap
Cc: Jens Axboe, linux-block, linux-kernel, Hans Holmberg (C)
On 01/05/2018 08:52 PM, Javier Gonzalez wrote:
>> On 5 Jan 2018, at 20.33, Randy Dunlap <rdunlap@infradead.org> wrote:
>>
>> On 01/05/2018 05:16 AM, Matias Bjørling wrote:
>>> From: Javier González <javier@cnexlabs.com>
>>>
>>> Allow to set the over-provision percentage on target creation. In case
>>> that the value is not provided, fall back to the default value set by
>>> the target.
>>>
>>> In pblk, set the default OP to 11% of the total size of the device
>>
>>> +#define PBLK_DEFAULT_OP (11)
>>
>> Hi,
>> Just curious -- where does 11 come from? Is it a spec value?
>
> 11 stands for 11% over-provisioning for the media to allow garbage
> collection. Different SSDs have different values based on the targeted
> workload - 11% is a common default value for standard drives.
>
> The spec. does not specify any default values of over-provisioning.
>
> Javier.
>
I think what Randy means is, why is the value not 20% or 7%, which is
the traditional over-provisioning on SSDs.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 19:53 ` Matias Bjørling
@ 2018-01-05 19:56 ` Javier Gonzalez
2018-01-05 20:17 ` Randy Dunlap
0 siblings, 1 reply; 45+ messages in thread
From: Javier Gonzalez @ 2018-01-05 19:56 UTC (permalink / raw)
To: Matias Bjørling
Cc: Randy Dunlap, Jens Axboe, linux-block, linux-kernel, Hans Holmberg (C)
[-- Attachment #1: Type: text/plain, Size: 1361 bytes --]
> On 5 Jan 2018, at 20.53, Matias Bjørling <m@bjorling.me> wrote:
>
> On 01/05/2018 08:52 PM, Javier Gonzalez wrote:
>>> On 5 Jan 2018, at 20.33, Randy Dunlap <rdunlap@infradead.org> wrote:
>>>
>>> On 01/05/2018 05:16 AM, Matias Bjørling wrote:
>>>> From: Javier González <javier@cnexlabs.com>
>>>>
>>>> Allow to set the over-provision percentage on target creation. In case
>>>> that the value is not provided, fall back to the default value set by
>>>> the target.
>>>>
>>>> In pblk, set the default OP to 11% of the total size of the device
>>>
>>>> +#define PBLK_DEFAULT_OP (11)
>>>
>>> Hi,
>>> Just curious -- where does 11 come from? Is it a spec value?
>> 11 stands for 11% over-provisioning for the media to allow garbage
>> collection. Different SSDs have different values based on the targeted
>> workload - 11% is a common default value for standard drives.
>> The spec. does not specify any default values of over-provisioning.
>> Javier.
>
> I think what Randy means is, why is the value not 20% or 7%, which is the traditional over-provisioning on SSDs.
We decided 11% based on customer input, but 7% and 20% are also good
default values. I'm ok with having any of those as default - anyone
caring about OP will define the value on target creation, which is the
primary objective of this patch.
Javier
[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl
2018-01-05 19:56 ` Javier Gonzalez
@ 2018-01-05 20:17 ` Randy Dunlap
0 siblings, 0 replies; 45+ messages in thread
From: Randy Dunlap @ 2018-01-05 20:17 UTC (permalink / raw)
To: Javier Gonzalez, Matias Bjørling
Cc: Jens Axboe, linux-block, linux-kernel, Hans Holmberg (C)
On 01/05/2018 11:56 AM, Javier Gonzalez wrote:
>> On 5 Jan 2018, at 20.53, Matias Bjørling <m@bjorling.me> wrote:
>>
>> On 01/05/2018 08:52 PM, Javier Gonzalez wrote:
>>>> On 5 Jan 2018, at 20.33, Randy Dunlap <rdunlap@infradead.org> wrote:
>>>>
>>>> On 01/05/2018 05:16 AM, Matias Bjørling wrote:
>>>>> From: Javier González <javier@cnexlabs.com>
>>>>>
>>>>> Allow to set the over-provision percentage on target creation. In case
>>>>> that the value is not provided, fall back to the default value set by
>>>>> the target.
>>>>>
>>>>> In pblk, set the default OP to 11% of the total size of the device
>>>>
>>>>> +#define PBLK_DEFAULT_OP (11)
>>>>
>>>> Hi,
>>>> Just curious -- where does 11 come from? Is it a spec value?
>>> 11 stands for 11% over-provisioning for the media to allow garbage
>>> collection. Different SSDs have different values based on the targeted
>>> workload - 11% is a common default value for standard drives.
>>> The spec. does not specify any default values of over-provisioning.
>>> Javier.
>>
>> I think what Randy means is, why is the value not 20% or 7%, which is the traditional over-provisioning on SSDs.
>
> We decided 11% based on customer input, but 7% and 20% are also good
> default values. I'm ok with having any of those as default - anyone
> caring about OP will define the value on target creation, which is the
> primary objective of this patch.
Got it. Thanks.
--
~Randy
^ permalink raw reply [flat|nested] 45+ messages in thread
* [GIT PULL 19/25] lightnvm: pblk: ignore high ecc errors on recovery
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (17 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 18/25] lightnvm: set target over-provision on create ioctl Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 20/25] lightnvm: pblk: do not log recovery read errors Matias Bjørling
` (6 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
On recovery, do not stop L2P recovery if reads report high ECC error
as the data is still available.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-recovery.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index fd38036..1d5e961 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -288,7 +288,7 @@ static int pblk_recov_read_oob(struct pblk *pblk, struct pblk_line *line,
/* At this point, the read should not fail. If it does, it is a problem
* we cannot recover from here. Need FTL log.
*/
- if (rqd->error) {
+ if (rqd->error && rqd->error != NVM_RSP_WARN_HIGHECC) {
pr_err("pblk: L2P recovery failed (%d)\n", rqd->error);
return -EINTR;
}
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 20/25] lightnvm: pblk: do not log recovery read errors
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (18 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 19/25] lightnvm: pblk: ignore high ecc errors on recovery Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 21/25] lightnvm: pblk: ensure kthread alloc. before kicking it Matias Bjørling
` (5 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
On scan recovery, reads can fail. This happens because the first page
for each line is read in order to determined if the line has been used
(and thus needs to be recovered), or not. This can lead to "empty page"
read errors.
Since these errors are normal, do not log them, as they are confusing
when reviewing the logs.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-core.c | 6 +++---
drivers/lightnvm/pblk.h | 1 +
2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/lightnvm/pblk-core.c b/drivers/lightnvm/pblk-core.c
index 5ec7ad6..0487b93 100644
--- a/drivers/lightnvm/pblk-core.c
+++ b/drivers/lightnvm/pblk-core.c
@@ -742,7 +742,7 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
cmd_op = NVM_OP_PWRITE;
flags = pblk_set_progr_mode(pblk, PBLK_WRITE);
lba_list = emeta_to_lbas(pblk, line->emeta->buf);
- } else if (dir == PBLK_READ) {
+ } else if (dir == PBLK_READ_RECOV || dir == PBLK_READ) {
bio_op = REQ_OP_READ;
cmd_op = NVM_OP_PREAD;
flags = pblk_set_read_mode(pblk, PBLK_READ_SEQUENTIAL);
@@ -802,7 +802,7 @@ static int pblk_line_submit_smeta_io(struct pblk *pblk, struct pblk_line *line,
if (rqd.error) {
if (dir == PBLK_WRITE)
pblk_log_write_err(pblk, &rqd);
- else
+ else if (dir == PBLK_READ)
pblk_log_read_err(pblk, &rqd);
}
@@ -816,7 +816,7 @@ int pblk_line_read_smeta(struct pblk *pblk, struct pblk_line *line)
{
u64 bpaddr = pblk_line_smeta_start(pblk, line);
- return pblk_line_submit_smeta_io(pblk, line, bpaddr, PBLK_READ);
+ return pblk_line_submit_smeta_io(pblk, line, bpaddr, PBLK_READ_RECOV);
}
int pblk_line_read_emeta(struct pblk *pblk, struct pblk_line *line,
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 19e622c..93ec4fd 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -60,6 +60,7 @@ enum {
PBLK_READ = READ,
PBLK_WRITE = WRITE,/* Write from write buffer */
PBLK_WRITE_INT, /* Internal write - no write buffer */
+ PBLK_READ_RECOV, /* Recovery read - errors allowed */
PBLK_ERASE,
};
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 21/25] lightnvm: pblk: ensure kthread alloc. before kicking it
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (19 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 20/25] lightnvm: pblk: do not log recovery read errors Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 22/25] lightnvm: pblk: free write buffer on init failure Matias Bjørling
` (4 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
When creating the write thread, ensure that the kthread has been created
before initializing the timer responsible from kicking it. Otherwise, if
the kthread creation fails or gets killed from used space, we risk
kicking an empty thread structure.
Also, since the kthread creation can be interrupted form user space,
adapt the error path to not report an error when this happens, since it
is intentional that the instance creation is aborted.
Signed-off-by: Javier González <javier@cnexlabs.com>
Updated source to reflect the new timer_setup API.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-init.c | 17 +++++++++++------
1 file changed, 11 insertions(+), 6 deletions(-)
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 533f690..7e11926 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -883,15 +883,19 @@ static int pblk_lines_init(struct pblk *pblk)
static int pblk_writer_init(struct pblk *pblk)
{
- timer_setup(&pblk->wtimer, pblk_write_timer_fn, 0);
- mod_timer(&pblk->wtimer, jiffies + msecs_to_jiffies(100));
-
pblk->writer_ts = kthread_create(pblk_write_ts, pblk, "pblk-writer-t");
if (IS_ERR(pblk->writer_ts)) {
- pr_err("pblk: could not allocate writer kthread\n");
- return PTR_ERR(pblk->writer_ts);
+ int err = PTR_ERR(pblk->writer_ts);
+
+ if (err != -EINTR)
+ pr_err("pblk: could not allocate writer kthread (%d)\n",
+ err);
+ return err;
}
+ timer_setup(&pblk->wtimer, pblk_write_timer_fn, 0);
+ mod_timer(&pblk->wtimer, jiffies + msecs_to_jiffies(100));
+
return 0;
}
@@ -1042,7 +1046,8 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
ret = pblk_writer_init(pblk);
if (ret) {
- pr_err("pblk: could not initialize write thread\n");
+ if (ret != -EINTR)
+ pr_err("pblk: could not initialize write thread\n");
goto fail_free_lines;
}
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 22/25] lightnvm: pblk: free write buffer on init failure
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (20 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 21/25] lightnvm: pblk: ensure kthread alloc. before kicking it Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 23/25] lightnvm: pblk: print instance name on instance info Matias Bjørling
` (3 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Refactor the way we free the write buffer to ensure that all entries get
freed in case of an error on the init sequence.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-init.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 7e11926..8c40bc3 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -355,6 +355,8 @@ static void pblk_core_free(struct pblk *pblk)
mempool_destroy(pblk->e_rq_pool);
mempool_destroy(pblk->w_rq_pool);
+ pblk_rwb_free(pblk);
+
pblk_free_global_caches(pblk);
}
@@ -931,7 +933,6 @@ static void pblk_tear_down(struct pblk *pblk)
pblk_pipeline_stop(pblk);
pblk_writer_stop(pblk);
pblk_rb_sync_l2p(&pblk->rwb);
- pblk_rwb_free(pblk);
pblk_rl_free(&pblk->rl);
pr_debug("pblk: consistent tear down\n");
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 23/25] lightnvm: pblk: print instance name on instance info
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (21 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 22/25] lightnvm: pblk: free write buffer on init failure Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 13:16 ` [GIT PULL 24/25] lightnvm: pblk: add iostat support Matias Bjørling
` (2 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Add the instance name to the information printed out on target creation.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-init.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index 8c40bc3..93d671c 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -1069,7 +1069,8 @@ static void *pblk_init(struct nvm_tgt_dev *dev, struct gendisk *tdisk,
blk_queue_max_discard_sectors(tqueue, UINT_MAX >> 9);
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, tqueue);
- pr_info("pblk init: luns:%u, lines:%d, secs:%llu, buf entries:%u\n",
+ pr_info("pblk(%s): luns:%u, lines:%d, secs:%llu, buf entries:%u\n",
+ tdisk->disk_name,
geo->all_luns, pblk->l_mg.nr_lines,
(unsigned long long)pblk->rl.nr_secs,
pblk->rwb.nr_entries);
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (22 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 23/25] lightnvm: pblk: print instance name on instance info Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 15:42 ` Jens Axboe
2018-01-05 13:16 ` [GIT PULL 25/25] lightnvm: pblk: refactor pblk_ppa_comp function Matias Bjørling
2018-01-05 15:50 ` Jens Axboe
25 siblings, 1 reply; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe
Cc: linux-block, linux-kernel, Javier González, Matias Bjørling
From: Javier González <javier@cnexlabs.com>
Since pblk registers its own block device, the iostat accounting is
not automatically done for us. Therefore, add the necessary
accounting logic to satisfy the iostat interface.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk-cache.c | 5 +++++
drivers/lightnvm/pblk-read.c | 31 +++++++++++++++++++------------
drivers/lightnvm/pblk.h | 1 +
3 files changed, 25 insertions(+), 12 deletions(-)
diff --git a/drivers/lightnvm/pblk-cache.c b/drivers/lightnvm/pblk-cache.c
index 0d227ef..000fcad 100644
--- a/drivers/lightnvm/pblk-cache.c
+++ b/drivers/lightnvm/pblk-cache.c
@@ -19,12 +19,16 @@
int pblk_write_to_cache(struct pblk *pblk, struct bio *bio, unsigned long flags)
{
+ struct request_queue *q = pblk->dev->q;
struct pblk_w_ctx w_ctx;
sector_t lba = pblk_get_lba(bio);
+ unsigned long start_time = jiffies;
unsigned int bpos, pos;
int nr_entries = pblk_get_secs(bio);
int i, ret;
+ generic_start_io_acct(q, WRITE, bio_sectors(bio), &pblk->disk->part0);
+
/* Update the write buffer head (mem) with the entries that we can
* write. The write in itself cannot fail, so there is no need to
* rollback from here on.
@@ -67,6 +71,7 @@ int pblk_write_to_cache(struct pblk *pblk, struct bio *bio, unsigned long flags)
pblk_rl_inserted(&pblk->rl, nr_entries);
out:
+ generic_end_io_acct(q, WRITE, &pblk->disk->part0, start_time);
pblk_write_should_kick(pblk);
return ret;
}
diff --git a/drivers/lightnvm/pblk-read.c b/drivers/lightnvm/pblk-read.c
index 0fe0c04..2f76128 100644
--- a/drivers/lightnvm/pblk-read.c
+++ b/drivers/lightnvm/pblk-read.c
@@ -158,8 +158,12 @@ static void pblk_end_user_read(struct bio *bio)
static void __pblk_end_io_read(struct pblk *pblk, struct nvm_rq *rqd,
bool put_line)
{
+ struct nvm_tgt_dev *dev = pblk->dev;
struct pblk_g_ctx *r_ctx = nvm_rq_to_pdu(rqd);
struct bio *bio = rqd->bio;
+ unsigned long start_time = r_ctx->start_time;
+
+ generic_end_io_acct(dev->q, READ, &pblk->disk->part0, start_time);
if (rqd->error)
pblk_log_read_err(pblk, rqd);
@@ -193,9 +197,9 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
__pblk_end_io_read(pblk, rqd, true);
}
-static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
- unsigned int bio_init_idx,
- unsigned long *read_bitmap)
+static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
+ unsigned int bio_init_idx,
+ unsigned long *read_bitmap)
{
struct bio *new_bio, *bio = rqd->bio;
struct pblk_sec_meta *meta_list = rqd->meta_list;
@@ -306,6 +310,8 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
return NVM_IO_OK;
err:
+ pr_err("pblk: failed to perform partial read\n");
+
/* Free allocated pages in new bio */
pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
__pblk_end_io_read(pblk, rqd, false);
@@ -357,6 +363,7 @@ static void pblk_read_rq(struct pblk *pblk, struct nvm_rq *rqd,
int pblk_submit_read(struct pblk *pblk, struct bio *bio)
{
struct nvm_tgt_dev *dev = pblk->dev;
+ struct request_queue *q = dev->q;
sector_t blba = pblk_get_lba(bio);
unsigned int nr_secs = pblk_get_secs(bio);
struct pblk_g_ctx *r_ctx;
@@ -372,6 +379,8 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
return NVM_IO_ERR;
}
+ generic_start_io_acct(q, READ, bio_sectors(bio), &pblk->disk->part0);
+
bitmap_zero(&read_bitmap, nr_secs);
rqd = pblk_alloc_rqd(pblk, PBLK_READ);
@@ -383,6 +392,7 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
rqd->end_io = pblk_end_io_read;
r_ctx = nvm_rq_to_pdu(rqd);
+ r_ctx->start_time = jiffies;
r_ctx->lba = blba;
/* Save the index for this bio's start. This is needed in case
@@ -422,7 +432,7 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
int_bio = bio_clone_fast(bio, GFP_KERNEL, pblk_bio_set);
if (!int_bio) {
pr_err("pblk: could not clone read bio\n");
- return NVM_IO_ERR;
+ goto fail_end_io;
}
rqd->bio = int_bio;
@@ -433,7 +443,7 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
pr_err("pblk: read IO submission failed\n");
if (int_bio)
bio_put(int_bio);
- return ret;
+ goto fail_end_io;
}
return NVM_IO_OK;
@@ -442,17 +452,14 @@ int pblk_submit_read(struct pblk *pblk, struct bio *bio)
/* The read bio request could be partially filled by the write buffer,
* but there are some holes that need to be read from the drive.
*/
- ret = pblk_fill_partial_read_bio(pblk, rqd, bio_init_idx, &read_bitmap);
- if (ret) {
- pr_err("pblk: failed to perform partial read\n");
- return ret;
- }
-
- return NVM_IO_OK;
+ return pblk_partial_read_bio(pblk, rqd, bio_init_idx, &read_bitmap);
fail_rqd_free:
pblk_free_rqd(pblk, rqd, PBLK_READ);
return ret;
+fail_end_io:
+ __pblk_end_io_read(pblk, rqd, false);
+ return ret;
}
static int read_ppalist_rq_gc(struct pblk *pblk, struct nvm_rq *rqd,
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 93ec4fd..8af374e 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -113,6 +113,7 @@ struct pblk_c_ctx {
/* read context */
struct pblk_g_ctx {
void *private;
+ unsigned long start_time;
u64 lba;
};
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-05 13:16 ` [GIT PULL 24/25] lightnvm: pblk: add iostat support Matias Bjørling
@ 2018-01-05 15:42 ` Jens Axboe
0 siblings, 0 replies; 45+ messages in thread
From: Jens Axboe @ 2018-01-05 15:42 UTC (permalink / raw)
To: Matias Bjørling; +Cc: linux-block, linux-kernel, Javier González
On Fri, Jan 05 2018, Matias Bj�rling wrote:
> From: Javier Gonz�lez <javier@cnexlabs.com>
>
> Since pblk registers its own block device, the iostat accounting is
> not automatically done for us. Therefore, add the necessary
> accounting logic to satisfy the iostat interface.
Ignorant question - why is it a raw block device, not using blk-mq?
> @@ -193,9 +197,9 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
> __pblk_end_io_read(pblk, rqd, true);
> }
>
> -static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> - unsigned int bio_init_idx,
> - unsigned long *read_bitmap)
> +static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> + unsigned int bio_init_idx,
> + unsigned long *read_bitmap)
> {
> struct bio *new_bio, *bio = rqd->bio;
> struct pblk_sec_meta *meta_list = rqd->meta_list;
> @@ -306,6 +310,8 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> return NVM_IO_OK;
>
> err:
> + pr_err("pblk: failed to perform partial read\n");
> +
> /* Free allocated pages in new bio */
> pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
> __pblk_end_io_read(pblk, rqd, false);
This seems to include unrelated changes, like the rename above and the
addition of the error logging?
--
Jens Axboe
^ permalink raw reply [flat|nested] 45+ messages in thread* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
@ 2018-01-05 15:42 ` Jens Axboe
0 siblings, 0 replies; 45+ messages in thread
From: Jens Axboe @ 2018-01-05 15:42 UTC (permalink / raw)
To: Matias Bjørling; +Cc: linux-block, linux-kernel, Javier González
On Fri, Jan 05 2018, Matias Bjørling wrote:
> From: Javier González <javier@cnexlabs.com>
>
> Since pblk registers its own block device, the iostat accounting is
> not automatically done for us. Therefore, add the necessary
> accounting logic to satisfy the iostat interface.
Ignorant question - why is it a raw block device, not using blk-mq?
> @@ -193,9 +197,9 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
> __pblk_end_io_read(pblk, rqd, true);
> }
>
> -static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> - unsigned int bio_init_idx,
> - unsigned long *read_bitmap)
> +static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> + unsigned int bio_init_idx,
> + unsigned long *read_bitmap)
> {
> struct bio *new_bio, *bio = rqd->bio;
> struct pblk_sec_meta *meta_list = rqd->meta_list;
> @@ -306,6 +310,8 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
> return NVM_IO_OK;
>
> err:
> + pr_err("pblk: failed to perform partial read\n");
> +
> /* Free allocated pages in new bio */
> pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
> __pblk_end_io_read(pblk, rqd, false);
This seems to include unrelated changes, like the rename above and the
addition of the error logging?
--
Jens Axboe
^ permalink raw reply [flat|nested] 45+ messages in thread* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-05 15:42 ` Jens Axboe
@ 2018-01-05 18:33 ` Matias Bjørling
-1 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 18:33 UTC (permalink / raw)
To: Jens Axboe; +Cc: linux-block, linux-kernel, Javier González
On 01/05/2018 04:42 PM, Jens Axboe wrote:
> On Fri, Jan 05 2018, Matias Bj�rling wrote:
>> From: Javier Gonz�lez <javier@cnexlabs.com>
>>
>> Since pblk registers its own block device, the iostat accounting is
>> not automatically done for us. Therefore, add the necessary
>> accounting logic to satisfy the iostat interface.
>
> Ignorant question - why is it a raw block device, not using blk-mq?
The current flow is using the raw block device, together with the blk-mq
nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path
in the /drivers/nvme/host/lightnvm.c file. From there it attaches the to
NVMe blk-mq implementation.
Is there a better way to do it?
>
>> @@ -193,9 +197,9 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
>> __pblk_end_io_read(pblk, rqd, true);
>> }
>>
>> -static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> - unsigned int bio_init_idx,
>> - unsigned long *read_bitmap)
>> +static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> + unsigned int bio_init_idx,
>> + unsigned long *read_bitmap)
>> {
>> struct bio *new_bio, *bio = rqd->bio;
>> struct pblk_sec_meta *meta_list = rqd->meta_list;
>> @@ -306,6 +310,8 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> return NVM_IO_OK;
>>
>> err:
>> + pr_err("pblk: failed to perform partial read\n");
>> +
>> /* Free allocated pages in new bio */
>> pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
>> __pblk_end_io_read(pblk, rqd, false);
>
> This seems to include unrelated changes, like the rename above and the
> addition of the error logging?
>
Grah... I missed it during review. It should have been its own patch or
part of the early refactor patches. Thanks for picking it up.
^ permalink raw reply [flat|nested] 45+ messages in thread* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
@ 2018-01-05 18:33 ` Matias Bjørling
0 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 18:33 UTC (permalink / raw)
To: Jens Axboe; +Cc: linux-block, linux-kernel, Javier González
On 01/05/2018 04:42 PM, Jens Axboe wrote:
> On Fri, Jan 05 2018, Matias Bjørling wrote:
>> From: Javier González <javier@cnexlabs.com>
>>
>> Since pblk registers its own block device, the iostat accounting is
>> not automatically done for us. Therefore, add the necessary
>> accounting logic to satisfy the iostat interface.
>
> Ignorant question - why is it a raw block device, not using blk-mq?
The current flow is using the raw block device, together with the blk-mq
nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path
in the /drivers/nvme/host/lightnvm.c file. From there it attaches the to
NVMe blk-mq implementation.
Is there a better way to do it?
>
>> @@ -193,9 +197,9 @@ static void pblk_end_io_read(struct nvm_rq *rqd)
>> __pblk_end_io_read(pblk, rqd, true);
>> }
>>
>> -static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> - unsigned int bio_init_idx,
>> - unsigned long *read_bitmap)
>> +static int pblk_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> + unsigned int bio_init_idx,
>> + unsigned long *read_bitmap)
>> {
>> struct bio *new_bio, *bio = rqd->bio;
>> struct pblk_sec_meta *meta_list = rqd->meta_list;
>> @@ -306,6 +310,8 @@ static int pblk_fill_partial_read_bio(struct pblk *pblk, struct nvm_rq *rqd,
>> return NVM_IO_OK;
>>
>> err:
>> + pr_err("pblk: failed to perform partial read\n");
>> +
>> /* Free allocated pages in new bio */
>> pblk_bio_free_pages(pblk, bio, 0, new_bio->bi_vcnt);
>> __pblk_end_io_read(pblk, rqd, false);
>
> This seems to include unrelated changes, like the rename above and the
> addition of the error logging?
>
Grah... I missed it during review. It should have been its own patch or
part of the early refactor patches. Thanks for picking it up.
^ permalink raw reply [flat|nested] 45+ messages in thread* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-05 18:33 ` Matias Bjørling
@ 2018-01-08 11:54 ` Christoph Hellwig
-1 siblings, 0 replies; 45+ messages in thread
From: Christoph Hellwig @ 2018-01-08 11:54 UTC (permalink / raw)
To: Matias Bjørling
Cc: Jens Axboe, linux-block, linux-kernel, Javier González
On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bj�rling wrote:
> On 01/05/2018 04:42 PM, Jens Axboe wrote:
> > On Fri, Jan 05 2018, Matias Bj�rling wrote:
> > > From: Javier Gonz�lez <javier@cnexlabs.com>
> > >
> > > Since pblk registers its own block device, the iostat accounting is
> > > not automatically done for us. Therefore, add the necessary
> > > accounting logic to satisfy the iostat interface.
> >
> > Ignorant question - why is it a raw block device, not using blk-mq?
>
> The current flow is using the raw block device, together with the blk-mq
> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path in
> the /drivers/nvme/host/lightnvm.c file. From there it attaches the to NVMe
> blk-mq implementation.
>
> Is there a better way to do it?
I suspect the right way to do things is to split NVMe for different
I/O command sets, and make this an I/O command set.
But before touching much of NVMe, I'd really, really like to see an
actual spec first.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
@ 2018-01-08 11:54 ` Christoph Hellwig
0 siblings, 0 replies; 45+ messages in thread
From: Christoph Hellwig @ 2018-01-08 11:54 UTC (permalink / raw)
To: Matias Bjørling
Cc: Jens Axboe, linux-block, linux-kernel, Javier González
On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bjørling wrote:
> On 01/05/2018 04:42 PM, Jens Axboe wrote:
> > On Fri, Jan 05 2018, Matias Bjørling wrote:
> > > From: Javier González <javier@cnexlabs.com>
> > >
> > > Since pblk registers its own block device, the iostat accounting is
> > > not automatically done for us. Therefore, add the necessary
> > > accounting logic to satisfy the iostat interface.
> >
> > Ignorant question - why is it a raw block device, not using blk-mq?
>
> The current flow is using the raw block device, together with the blk-mq
> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path in
> the /drivers/nvme/host/lightnvm.c file. From there it attaches the to NVMe
> blk-mq implementation.
>
> Is there a better way to do it?
I suspect the right way to do things is to split NVMe for different
I/O command sets, and make this an I/O command set.
But before touching much of NVMe, I'd really, really like to see an
actual spec first.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-08 11:54 ` Christoph Hellwig
@ 2018-01-08 12:53 ` Javier González
-1 siblings, 0 replies; 45+ messages in thread
From: Javier González @ 2018-01-08 12:53 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Matias Bjørling, Jens Axboe, linux-block, linux-kernel
> On 8 Jan 2018, at 12.54, Christoph Hellwig <hch@infradead.org> wrote:
>=20
> On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bj=C3=B8rling wrote:
>> On 01/05/2018 04:42 PM, Jens Axboe wrote:
>>> On Fri, Jan 05 2018, Matias Bj=C3=B8rling wrote:
>>>> From: Javier Gonz=C3=A1lez <javier@cnexlabs.com>
>>>>=20
>>>> Since pblk registers its own block device, the iostat accounting is
>>>> not automatically done for us. Therefore, add the necessary
>>>> accounting logic to satisfy the iostat interface.
>>>=20
>>> Ignorant question - why is it a raw block device, not using blk-mq?
>>=20
>> The current flow is using the raw block device, together with the =
blk-mq
>> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() =
path in
>> the /drivers/nvme/host/lightnvm.c file. =46rom there it attaches the =
to NVMe
>> blk-mq implementation.
>>=20
>> Is there a better way to do it?
>=20
> I suspect the right way to do things is to split NVMe for different
> I/O command sets, and make this an I/O command set.
This makes sense. This was actually how I implemented it to start with,
but I changed it to be less intrusive on the nvme path. Let's revert the
patch and we can add it back when we push the 2.0 patches.
> But before touching much of NVMe, I'd really, really like to see an
> actual spec first.
The 2.0 spec. is open and is available here [1]. I thought you had
looked into it already... Anyway, feedback is more than welcome.
[1] =
https://docs.google.com/document/d/1kedBY_1-hfkAlqT4EdwY6gz-6UOZbn7kIjWpmB=
LPNj0
Javier=
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
@ 2018-01-08 12:53 ` Javier González
0 siblings, 0 replies; 45+ messages in thread
From: Javier González @ 2018-01-08 12:53 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Matias Bjørling, Jens Axboe, linux-block, linux-kernel
> On 8 Jan 2018, at 12.54, Christoph Hellwig <hch@infradead.org> wrote:
>
> On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bjørling wrote:
>> On 01/05/2018 04:42 PM, Jens Axboe wrote:
>>> On Fri, Jan 05 2018, Matias Bjørling wrote:
>>>> From: Javier González <javier@cnexlabs.com>
>>>>
>>>> Since pblk registers its own block device, the iostat accounting is
>>>> not automatically done for us. Therefore, add the necessary
>>>> accounting logic to satisfy the iostat interface.
>>>
>>> Ignorant question - why is it a raw block device, not using blk-mq?
>>
>> The current flow is using the raw block device, together with the blk-mq
>> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path in
>> the /drivers/nvme/host/lightnvm.c file. From there it attaches the to NVMe
>> blk-mq implementation.
>>
>> Is there a better way to do it?
>
> I suspect the right way to do things is to split NVMe for different
> I/O command sets, and make this an I/O command set.
This makes sense. This was actually how I implemented it to start with,
but I changed it to be less intrusive on the nvme path. Let's revert the
patch and we can add it back when we push the 2.0 patches.
> But before touching much of NVMe, I'd really, really like to see an
> actual spec first.
The 2.0 spec. is open and is available here [1]. I thought you had
looked into it already... Anyway, feedback is more than welcome.
[1] https://docs.google.com/document/d/1kedBY_1-hfkAlqT4EdwY6gz-6UOZbn7kIjWpmBLPNj0
Javier
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
2018-01-08 12:53 ` Javier González
@ 2018-01-08 13:31 ` Matias Bjørling
-1 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-08 13:31 UTC (permalink / raw)
To: Javier González; +Cc: Christoph Hellwig, Jens Axboe, linux-block, LKML
On Mon, Jan 8, 2018 at 1:53 PM, Javier Gonz=C3=A1lez <jg@lightnvm.io> wrote=
:
>> On 8 Jan 2018, at 12.54, Christoph Hellwig <hch@infradead.org> wrote:
>>
>> On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bj=C3=B8rling wrote:
>>> On 01/05/2018 04:42 PM, Jens Axboe wrote:
>>>> On Fri, Jan 05 2018, Matias Bj=C3=B8rling wrote:
>>>>> From: Javier Gonz=C3=A1lez <javier@cnexlabs.com>
>>>>>
>>>>> Since pblk registers its own block device, the iostat accounting is
>>>>> not automatically done for us. Therefore, add the necessary
>>>>> accounting logic to satisfy the iostat interface.
>>>>
>>>> Ignorant question - why is it a raw block device, not using blk-mq?
>>>
>>> The current flow is using the raw block device, together with the blk-m=
q
>>> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path=
in
>>> the /drivers/nvme/host/lightnvm.c file. From there it attaches the to N=
VMe
>>> blk-mq implementation.
>>>
>>> Is there a better way to do it?
>>
>> I suspect the right way to do things is to split NVMe for different
>> I/O command sets, and make this an I/O command set.
>
> This makes sense. This was actually how I implemented it to start with,
> but I changed it to be less intrusive on the nvme path. Let's revert the
> patch and we can add it back when we push the 2.0 patches.
>
>> But before touching much of NVMe, I'd really, really like to see an
>> actual spec first.
>
> The 2.0 spec. is open and is available here [1]. I thought you had
> looked into it already... Anyway, feedback is more than welcome.
>
> [1] https://docs.google.com/document/d/1kedBY_1-hfkAlqT4EdwY6gz-6UOZbn7kI=
jWpmBLPNj0
>
> Javier
The 2.0 spec is still under development. No reason to redo the I/O
stacks until it is final.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 24/25] lightnvm: pblk: add iostat support
@ 2018-01-08 13:31 ` Matias Bjørling
0 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-08 13:31 UTC (permalink / raw)
To: Javier González; +Cc: Christoph Hellwig, Jens Axboe, linux-block, LKML
On Mon, Jan 8, 2018 at 1:53 PM, Javier González <jg@lightnvm.io> wrote:
>> On 8 Jan 2018, at 12.54, Christoph Hellwig <hch@infradead.org> wrote:
>>
>> On Fri, Jan 05, 2018 at 07:33:36PM +0100, Matias Bjørling wrote:
>>> On 01/05/2018 04:42 PM, Jens Axboe wrote:
>>>> On Fri, Jan 05 2018, Matias Bjørling wrote:
>>>>> From: Javier González <javier@cnexlabs.com>
>>>>>
>>>>> Since pblk registers its own block device, the iostat accounting is
>>>>> not automatically done for us. Therefore, add the necessary
>>>>> accounting logic to satisfy the iostat interface.
>>>>
>>>> Ignorant question - why is it a raw block device, not using blk-mq?
>>>
>>> The current flow is using the raw block device, together with the blk-mq
>>> nvme device driver. A bio is sent down to the nvme_nvm_submit_io() path in
>>> the /drivers/nvme/host/lightnvm.c file. From there it attaches the to NVMe
>>> blk-mq implementation.
>>>
>>> Is there a better way to do it?
>>
>> I suspect the right way to do things is to split NVMe for different
>> I/O command sets, and make this an I/O command set.
>
> This makes sense. This was actually how I implemented it to start with,
> but I changed it to be less intrusive on the nvme path. Let's revert the
> patch and we can add it back when we push the 2.0 patches.
>
>> But before touching much of NVMe, I'd really, really like to see an
>> actual spec first.
>
> The 2.0 spec. is open and is available here [1]. I thought you had
> looked into it already... Anyway, feedback is more than welcome.
>
> [1] https://docs.google.com/document/d/1kedBY_1-hfkAlqT4EdwY6gz-6UOZbn7kIjWpmBLPNj0
>
> Javier
The 2.0 spec is still under development. No reason to redo the I/O
stacks until it is final.
^ permalink raw reply [flat|nested] 45+ messages in thread
* [GIT PULL 25/25] lightnvm: pblk: refactor pblk_ppa_comp function
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
` (23 preceding siblings ...)
2018-01-05 13:16 ` [GIT PULL 24/25] lightnvm: pblk: add iostat support Matias Bjørling
@ 2018-01-05 13:16 ` Matias Bjørling
2018-01-05 15:50 ` Jens Axboe
25 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 13:16 UTC (permalink / raw)
To: axboe; +Cc: linux-block, linux-kernel, Matias Bjørling
Shorten function to simply return the value of the if statement.
Signed-off-by: Matias Bjørling <m@bjorling.me>
---
drivers/lightnvm/pblk.h | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 8af374e..8c357fb 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -1047,10 +1047,7 @@ static inline void pblk_ppa_set_empty(struct ppa_addr *ppa_addr)
static inline bool pblk_ppa_comp(struct ppa_addr lppa, struct ppa_addr rppa)
{
- if (lppa.ppa == rppa.ppa)
- return true;
-
- return false;
+ return (lppa.ppa == rppa.ppa);
}
static inline int pblk_addr_in_cache(struct ppa_addr ppa)
--
2.9.3
^ permalink raw reply related [flat|nested] 45+ messages in thread* Re: [GIT PULL 00/25] LightNVM updates for 4.16
2018-01-05 13:15 [GIT PULL 00/25] LightNVM updates for 4.16 Matias Bjørling
@ 2018-01-05 15:50 ` Jens Axboe
2018-01-05 13:15 ` [GIT PULL 02/25] lightnvm: remove rrpc Matias Bjørling
` (24 subsequent siblings)
25 siblings, 0 replies; 45+ messages in thread
From: Jens Axboe @ 2018-01-05 15:50 UTC (permalink / raw)
To: Matias Bjørling; +Cc: linux-block, linux-kernel
On Fri, Jan 05 2018, Matias Bj�rling wrote:
> Hi Jens,
>
> Here is a couple of patches for 4.16.
>
> This patchset prepares the lightnvm and pblk source code for the 2.0
> specification release. The specification is close to its final
> revision. After these changes, 2.0 support is a quick drop.
>
> While adopting 2.0 specification, the rrpc and null_blk
> implementation had to go. rrpc used the deprecated device-side
> table mapping table API for the 1.2 specification, and there are
> no users of it. Similarly, null_blk has not worked with pblk for
> a while.
>
> Javier and Hans have been hard at work at fixing up bugs in pblk,
> especially concerning synchronization points between user I/O and GC
> I/O. While also fixing bugs that occurred on bringup/bringdown.
>
> They also added support for iostats and added an option to add
> support for defining the amount of over-provisioning an instance
> should have.
Apart from the minor comment on one of the later patches, looks fine to
me. Applied for 4.16.
--
Jens Axboe
^ permalink raw reply [flat|nested] 45+ messages in thread* Re: [GIT PULL 00/25] LightNVM updates for 4.16
@ 2018-01-05 15:50 ` Jens Axboe
0 siblings, 0 replies; 45+ messages in thread
From: Jens Axboe @ 2018-01-05 15:50 UTC (permalink / raw)
To: Matias Bjørling; +Cc: linux-block, linux-kernel
On Fri, Jan 05 2018, Matias Bjørling wrote:
> Hi Jens,
>
> Here is a couple of patches for 4.16.
>
> This patchset prepares the lightnvm and pblk source code for the 2.0
> specification release. The specification is close to its final
> revision. After these changes, 2.0 support is a quick drop.
>
> While adopting 2.0 specification, the rrpc and null_blk
> implementation had to go. rrpc used the deprecated device-side
> table mapping table API for the 1.2 specification, and there are
> no users of it. Similarly, null_blk has not worked with pblk for
> a while.
>
> Javier and Hans have been hard at work at fixing up bugs in pblk,
> especially concerning synchronization points between user I/O and GC
> I/O. While also fixing bugs that occurred on bringup/bringdown.
>
> They also added support for iostats and added an option to add
> support for defining the amount of over-provisioning an instance
> should have.
Apart from the minor comment on one of the later patches, looks fine to
me. Applied for 4.16.
--
Jens Axboe
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 00/25] LightNVM updates for 4.16
2018-01-05 15:50 ` Jens Axboe
@ 2018-01-05 18:34 ` Matias Bjørling
-1 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 18:34 UTC (permalink / raw)
To: Jens Axboe; +Cc: linux-block, linux-kernel
On 01/05/2018 04:50 PM, Jens Axboe wrote:
> On Fri, Jan 05 2018, Matias Bj�rling wrote:
>> Hi Jens,
>>
>> Here is a couple of patches for 4.16.
>>
>> This patchset prepares the lightnvm and pblk source code for the 2.0
>> specification release. The specification is close to its final
>> revision. After these changes, 2.0 support is a quick drop.
>>
>> While adopting 2.0 specification, the rrpc and null_blk
>> implementation had to go. rrpc used the deprecated device-side
>> table mapping table API for the 1.2 specification, and there are
>> no users of it. Similarly, null_blk has not worked with pblk for
>> a while.
>>
>> Javier and Hans have been hard at work at fixing up bugs in pblk,
>> especially concerning synchronization points between user I/O and GC
>> I/O. While also fixing bugs that occurred on bringup/bringdown.
>>
>> They also added support for iostats and added an option to add
>> support for defining the amount of over-provisioning an instance
>> should have.
>
> Apart from the minor comment on one of the later patches, looks fine to
> me. Applied for 4.16.
>
Thanks Jens.
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [GIT PULL 00/25] LightNVM updates for 4.16
@ 2018-01-05 18:34 ` Matias Bjørling
0 siblings, 0 replies; 45+ messages in thread
From: Matias Bjørling @ 2018-01-05 18:34 UTC (permalink / raw)
To: Jens Axboe; +Cc: linux-block, linux-kernel
On 01/05/2018 04:50 PM, Jens Axboe wrote:
> On Fri, Jan 05 2018, Matias Bjørling wrote:
>> Hi Jens,
>>
>> Here is a couple of patches for 4.16.
>>
>> This patchset prepares the lightnvm and pblk source code for the 2.0
>> specification release. The specification is close to its final
>> revision. After these changes, 2.0 support is a quick drop.
>>
>> While adopting 2.0 specification, the rrpc and null_blk
>> implementation had to go. rrpc used the deprecated device-side
>> table mapping table API for the 1.2 specification, and there are
>> no users of it. Similarly, null_blk has not worked with pblk for
>> a while.
>>
>> Javier and Hans have been hard at work at fixing up bugs in pblk,
>> especially concerning synchronization points between user I/O and GC
>> I/O. While also fixing bugs that occurred on bringup/bringdown.
>>
>> They also added support for iostats and added an option to add
>> support for defining the amount of over-provisioning an instance
>> should have.
>
> Apart from the minor comment on one of the later patches, looks fine to
> me. Applied for 4.16.
>
Thanks Jens.
^ permalink raw reply [flat|nested] 45+ messages in thread