@@ -115,6 +115,18 @@ module_param_string(init_hctx, g_init_hctx_str, sizeof(g_init_hctx_str), 0444);
MODULE_PARM_DESC(init_hctx, "Fault injection to fail hctx init. init_hctx=<interval>,<probability>,<space>,<times>");
#endif
+/*
+ * Historic queue modes.
+ *
+ * These days nothing but NULL_Q_MQ is actually supported, but we keep it the
+ * enum for error reporting.
+ */
+enum {
+ NULL_Q_BIO = 0,
+ NULL_Q_RQ = 1,
+ NULL_Q_MQ = 2,
+};
+
static int g_queue_mode = NULL_Q_MQ;
static int null_param_store_val(const char *str, int *val, int min, int max)
@@ -756,98 +768,11 @@ static void null_free_dev(struct nullb_device *dev)
kfree(dev);
}
-static void put_tag(struct nullb_queue *nq, unsigned int tag)
-{
- clear_bit_unlock(tag, nq->tag_map);
-
- if (waitqueue_active(&nq->wait))
- wake_up(&nq->wait);
-}
-
-static unsigned int get_tag(struct nullb_queue *nq)
-{
- unsigned int tag;
-
- do {
- tag = find_first_zero_bit(nq->tag_map, nq->queue_depth);
- if (tag >= nq->queue_depth)
- return -1U;
- } while (test_and_set_bit_lock(tag, nq->tag_map));
-
- return tag;
-}
-
-static void free_cmd(struct nullb_cmd *cmd)
-{
- put_tag(cmd->nq, cmd->tag);
-}
-
-static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer);
-
-static struct nullb_cmd *__alloc_cmd(struct nullb_queue *nq)
-{
- struct nullb_cmd *cmd;
- unsigned int tag;
-
- tag = get_tag(nq);
- if (tag != -1U) {
- cmd = &nq->cmds[tag];
- cmd->tag = tag;
- cmd->error = BLK_STS_OK;
- cmd->nq = nq;
- if (nq->dev->irqmode == NULL_IRQ_TIMER) {
- hrtimer_init(&cmd->timer, CLOCK_MONOTONIC,
- HRTIMER_MODE_REL);
- cmd->timer.function = null_cmd_timer_expired;
- }
- return cmd;
- }
-
- return NULL;
-}
-
-static struct nullb_cmd *alloc_cmd(struct nullb_queue *nq, struct bio *bio)
-{
- struct nullb_cmd *cmd;
- DEFINE_WAIT(wait);
-
- do {
- /*
- * This avoids multiple return statements, multiple calls to
- * __alloc_cmd() and a fast path call to prepare_to_wait().
- */
- cmd = __alloc_cmd(nq);
- if (cmd) {
- cmd->bio = bio;
- return cmd;
- }
- prepare_to_wait(&nq->wait, &wait, TASK_UNINTERRUPTIBLE);
- io_schedule();
- finish_wait(&nq->wait, &wait);
- } while (1);
-}
-
-static void end_cmd(struct nullb_cmd *cmd)
-{
- int queue_mode = cmd->nq->dev->queue_mode;
-
- switch (queue_mode) {
- case NULL_Q_MQ:
- blk_mq_end_request(cmd->rq, cmd->error);
- return;
- case NULL_Q_BIO:
- cmd->bio->bi_status = cmd->error;
- bio_endio(cmd->bio);
- break;
- }
-
- free_cmd(cmd);
-}
-
static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
{
- end_cmd(container_of(timer, struct nullb_cmd, timer));
+ struct nullb_cmd *cmd = container_of(timer, struct nullb_cmd, timer);
+ blk_mq_end_request(blk_mq_rq_from_pdu(cmd), cmd->error);
return HRTIMER_NORESTART;
}
@@ -860,7 +785,9 @@ static void null_cmd_end_timer(struct nullb_cmd *cmd)
static void null_complete_rq(struct request *rq)
{
- end_cmd(blk_mq_rq_to_pdu(rq));
+ struct nullb_cmd *cmd = blk_mq_rq_to_pdu(rq);
+
+ blk_mq_end_request(rq, cmd->error);
}
static struct nullb_page *null_alloc_page(void)
@@ -1277,7 +1204,7 @@ static int null_transfer(struct nullb *nullb, struct page *page,
static int null_handle_rq(struct nullb_cmd *cmd)
{
- struct request *rq = cmd->rq;
+ struct request *rq = blk_mq_rq_from_pdu(cmd);
struct nullb *nullb = cmd->nq->dev->nullb;
int err;
unsigned int len;
@@ -1302,63 +1229,21 @@ static int null_handle_rq(struct nullb_cmd *cmd)
return 0;
}
-static int null_handle_bio(struct nullb_cmd *cmd)
-{
- struct bio *bio = cmd->bio;
- struct nullb *nullb = cmd->nq->dev->nullb;
- int err;
- unsigned int len;
- sector_t sector = bio->bi_iter.bi_sector;
- struct bio_vec bvec;
- struct bvec_iter iter;
-
- spin_lock_irq(&nullb->lock);
- bio_for_each_segment(bvec, bio, iter) {
- len = bvec.bv_len;
- err = null_transfer(nullb, bvec.bv_page, len, bvec.bv_offset,
- op_is_write(bio_op(bio)), sector,
- bio->bi_opf & REQ_FUA);
- if (err) {
- spin_unlock_irq(&nullb->lock);
- return err;
- }
- sector += len >> SECTOR_SHIFT;
- }
- spin_unlock_irq(&nullb->lock);
- return 0;
-}
-
-static void null_stop_queue(struct nullb *nullb)
-{
- struct request_queue *q = nullb->q;
-
- if (nullb->dev->queue_mode == NULL_Q_MQ)
- blk_mq_stop_hw_queues(q);
-}
-
-static void null_restart_queue_async(struct nullb *nullb)
-{
- struct request_queue *q = nullb->q;
-
- if (nullb->dev->queue_mode == NULL_Q_MQ)
- blk_mq_start_stopped_hw_queues(q, true);
-}
-
static inline blk_status_t null_handle_throttled(struct nullb_cmd *cmd)
{
struct nullb_device *dev = cmd->nq->dev;
struct nullb *nullb = dev->nullb;
blk_status_t sts = BLK_STS_OK;
- struct request *rq = cmd->rq;
+ struct request *rq = blk_mq_rq_from_pdu(cmd);
if (!hrtimer_active(&nullb->bw_timer))
hrtimer_restart(&nullb->bw_timer);
if (atomic_long_sub_return(blk_rq_bytes(rq), &nullb->cur_bytes) < 0) {
- null_stop_queue(nullb);
+ blk_mq_stop_hw_queues(nullb->q);
/* race with timer */
if (atomic_long_read(&nullb->cur_bytes) > 0)
- null_restart_queue_async(nullb);
+ blk_mq_start_stopped_hw_queues(nullb->q, true);
/* requeue request */
sts = BLK_STS_DEV_RESOURCE;
}
@@ -1385,37 +1270,29 @@ static inline blk_status_t null_handle_memory_backed(struct nullb_cmd *cmd,
sector_t nr_sectors)
{
struct nullb_device *dev = cmd->nq->dev;
- int err;
if (op == REQ_OP_DISCARD)
return null_handle_discard(dev, sector, nr_sectors);
+ return errno_to_blk_status(null_handle_rq(cmd));
- if (dev->queue_mode == NULL_Q_BIO)
- err = null_handle_bio(cmd);
- else
- err = null_handle_rq(cmd);
-
- return errno_to_blk_status(err);
}
static void nullb_zero_read_cmd_buffer(struct nullb_cmd *cmd)
{
+ struct request *rq = blk_mq_rq_from_pdu(cmd);
struct nullb_device *dev = cmd->nq->dev;
struct bio *bio;
- if (dev->memory_backed)
- return;
-
- if (dev->queue_mode == NULL_Q_BIO && bio_op(cmd->bio) == REQ_OP_READ) {
- zero_fill_bio(cmd->bio);
- } else if (req_op(cmd->rq) == REQ_OP_READ) {
- __rq_for_each_bio(bio, cmd->rq)
+ if (!dev->memory_backed && req_op(rq) == REQ_OP_READ) {
+ __rq_for_each_bio(bio, rq)
zero_fill_bio(bio);
}
}
static inline void nullb_complete_cmd(struct nullb_cmd *cmd)
{
+ struct request *rq = blk_mq_rq_from_pdu(cmd);
+
/*
* Since root privileges are required to configure the null_blk
* driver, it is fine that this driver does not initialize the
@@ -1429,20 +1306,10 @@ static inline void nullb_complete_cmd(struct nullb_cmd *cmd)
/* Complete IO by inline, softirq or timer */
switch (cmd->nq->dev->irqmode) {
case NULL_IRQ_SOFTIRQ:
- switch (cmd->nq->dev->queue_mode) {
- case NULL_Q_MQ:
- blk_mq_complete_request(cmd->rq);
- break;
- case NULL_Q_BIO:
- /*
- * XXX: no proper submitting cpu information available.
- */
- end_cmd(cmd);
- break;
- }
+ blk_mq_complete_request(rq);
break;
case NULL_IRQ_NONE:
- end_cmd(cmd);
+ blk_mq_end_request(rq, cmd->error);
break;
case NULL_IRQ_TIMER:
null_cmd_end_timer(cmd);
@@ -1503,7 +1370,7 @@ static enum hrtimer_restart nullb_bwtimer_fn(struct hrtimer *timer)
return HRTIMER_NORESTART;
atomic_long_set(&nullb->cur_bytes, mb_per_tick(mbps));
- null_restart_queue_async(nullb);
+ blk_mq_start_stopped_hw_queues(nullb->q, true);
hrtimer_forward_now(&nullb->bw_timer, timer_interval);
@@ -1520,26 +1387,6 @@ static void nullb_setup_bwtimer(struct nullb *nullb)
hrtimer_start(&nullb->bw_timer, timer_interval, HRTIMER_MODE_REL);
}
-static struct nullb_queue *nullb_to_queue(struct nullb *nullb)
-{
- int index = 0;
-
- if (nullb->nr_queues != 1)
- index = raw_smp_processor_id() / ((nr_cpu_ids + nullb->nr_queues - 1) / nullb->nr_queues);
-
- return &nullb->queues[index];
-}
-
-static void null_submit_bio(struct bio *bio)
-{
- sector_t sector = bio->bi_iter.bi_sector;
- sector_t nr_sectors = bio_sectors(bio);
- struct nullb *nullb = bio->bi_bdev->bd_disk->private_data;
- struct nullb_queue *nq = nullb_to_queue(nullb);
-
- null_handle_cmd(alloc_cmd(nq, bio), sector, nr_sectors, bio_op(bio));
-}
-
#ifdef CONFIG_BLK_DEV_NULL_BLK_FAULT_INJECTION
static bool should_timeout_request(struct request *rq)
@@ -1659,7 +1506,7 @@ static int null_poll(struct blk_mq_hw_ctx *hctx, struct io_comp_batch *iob)
blk_rq_sectors(req));
if (!blk_mq_add_to_batch(req, iob, (__force int) cmd->error,
blk_mq_end_request_batch))
- end_cmd(cmd);
+ blk_mq_end_request(req, cmd->error);
nr++;
}
@@ -1715,7 +1562,6 @@ static blk_status_t null_queue_rq(struct blk_mq_hw_ctx *hctx,
hrtimer_init(&cmd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
cmd->timer.function = null_cmd_timer_expired;
}
- cmd->rq = rq;
cmd->error = BLK_STS_OK;
cmd->nq = nq;
cmd->fake_timeout = should_timeout_request(rq) ||
@@ -1774,22 +1620,6 @@ static void null_queue_rqs(struct request **rqlist)
*rqlist = requeue_list;
}
-static void cleanup_queue(struct nullb_queue *nq)
-{
- bitmap_free(nq->tag_map);
- kfree(nq->cmds);
-}
-
-static void cleanup_queues(struct nullb *nullb)
-{
- int i;
-
- for (i = 0; i < nullb->nr_queues; i++)
- cleanup_queue(&nullb->queues[i]);
-
- kfree(nullb->queues);
-}
-
static void null_exit_hctx(struct blk_mq_hw_ctx *hctx, unsigned int hctx_idx)
{
struct nullb_queue *nq = hctx->driver_data;
@@ -1800,8 +1630,6 @@ static void null_exit_hctx(struct blk_mq_hw_ctx *hctx, unsigned int hctx_idx)
static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq)
{
- init_waitqueue_head(&nq->wait);
- nq->queue_depth = nullb->queue_depth;
nq->dev = nullb->dev;
INIT_LIST_HEAD(&nq->poll_list);
spin_lock_init(&nq->poll_lock);
@@ -1853,14 +1681,13 @@ static void null_del_dev(struct nullb *nullb)
if (test_bit(NULLB_DEV_FL_THROTTLED, &nullb->dev->flags)) {
hrtimer_cancel(&nullb->bw_timer);
atomic_long_set(&nullb->cur_bytes, LONG_MAX);
- null_restart_queue_async(nullb);
+ blk_mq_start_stopped_hw_queues(nullb->q, true);
}
put_disk(nullb->disk);
- if (dev->queue_mode == NULL_Q_MQ &&
- nullb->tag_set == &nullb->__tag_set)
+ if (nullb->tag_set == &nullb->__tag_set)
blk_mq_free_tag_set(nullb->tag_set);
- cleanup_queues(nullb);
+ kfree(nullb->queues);
if (null_cache_active(nullb))
null_free_device_storage(nullb->dev, true);
kfree(nullb);
@@ -1887,40 +1714,11 @@ static void null_config_discard(struct nullb *nullb)
blk_queue_max_discard_sectors(nullb->q, UINT_MAX >> 9);
}
-static const struct block_device_operations null_bio_ops = {
+static const struct block_device_operations null_ops = {
.owner = THIS_MODULE,
- .submit_bio = null_submit_bio,
.report_zones = null_report_zones,
};
-static const struct block_device_operations null_rq_ops = {
- .owner = THIS_MODULE,
- .report_zones = null_report_zones,
-};
-
-static int setup_commands(struct nullb_queue *nq)
-{
- struct nullb_cmd *cmd;
- int i;
-
- nq->cmds = kcalloc(nq->queue_depth, sizeof(*cmd), GFP_KERNEL);
- if (!nq->cmds)
- return -ENOMEM;
-
- nq->tag_map = bitmap_zalloc(nq->queue_depth, GFP_KERNEL);
- if (!nq->tag_map) {
- kfree(nq->cmds);
- return -ENOMEM;
- }
-
- for (i = 0; i < nq->queue_depth; i++) {
- cmd = &nq->cmds[i];
- cmd->tag = -1U;
- }
-
- return 0;
-}
-
static int setup_queues(struct nullb *nullb)
{
int nqueues = nr_cpu_ids;
@@ -1937,24 +1735,6 @@ static int setup_queues(struct nullb *nullb)
return 0;
}
-static int init_driver_queues(struct nullb *nullb)
-{
- struct nullb_queue *nq;
- int i, ret = 0;
-
- for (i = 0; i < nullb->dev->submit_queues; i++) {
- nq = &nullb->queues[i];
-
- null_init_queue(nullb, nq);
-
- ret = setup_commands(nq);
- if (ret)
- return ret;
- nullb->nr_queues++;
- }
- return 0;
-}
-
static int null_gendisk_register(struct nullb *nullb)
{
sector_t size = ((sector_t)nullb->dev->size * SZ_1M) >> SECTOR_SHIFT;
@@ -1965,10 +1745,7 @@ static int null_gendisk_register(struct nullb *nullb)
disk->major = null_major;
disk->first_minor = nullb->index;
disk->minors = 1;
- if (queue_is_mq(nullb->q))
- disk->fops = &null_rq_ops;
- else
- disk->fops = &null_bio_ops;
+ disk->fops = &null_ops;
disk->private_data = nullb;
strscpy_pad(disk->disk_name, nullb->disk_name, DISK_NAME_LEN);
@@ -2036,11 +1813,15 @@ static int null_validate_conf(struct nullb_device *dev)
pr_err("legacy IO path is no longer available\n");
return -EINVAL;
}
+ if (dev->queue_mode == NULL_Q_BIO) {
+ pr_err("BIO-based IO path is no longer available, using blk-mq instead.\n");
+ dev->queue_mode = NULL_Q_MQ;
+ }
dev->blocksize = round_down(dev->blocksize, 512);
dev->blocksize = clamp_t(unsigned int, dev->blocksize, 512, 4096);
- if (dev->queue_mode == NULL_Q_MQ && dev->use_per_node_hctx) {
+ if (dev->use_per_node_hctx) {
if (dev->submit_queues != nr_online_nodes)
dev->submit_queues = nr_online_nodes;
} else if (dev->submit_queues > nr_cpu_ids)
@@ -2052,8 +1833,6 @@ static int null_validate_conf(struct nullb_device *dev)
if (dev->poll_queues > g_poll_queues)
dev->poll_queues = g_poll_queues;
dev->prev_poll_queues = dev->poll_queues;
-
- dev->queue_mode = min_t(unsigned int, dev->queue_mode, NULL_Q_MQ);
dev->irqmode = min_t(unsigned int, dev->irqmode, NULL_IRQ_TIMER);
/* Do memory allocation, so set blocking */
@@ -2064,9 +1843,6 @@ static int null_validate_conf(struct nullb_device *dev)
dev->cache_size = min_t(unsigned long, ULONG_MAX / 1024 / 1024,
dev->cache_size);
dev->mbps = min_t(unsigned int, 1024 * 40, dev->mbps);
- /* can not stop a queue */
- if (dev->queue_mode == NULL_Q_BIO)
- dev->mbps = 0;
if (dev->zoned &&
(!dev->zone_size || !is_power_of_2(dev->zone_size))) {
@@ -2127,44 +1903,31 @@ static int null_add_dev(struct nullb_device *dev)
if (rv)
goto out_free_nullb;
- if (dev->queue_mode == NULL_Q_MQ) {
- if (dev->shared_tags) {
- if (!tag_set.ops) {
- rv = null_init_tag_set(NULL, &tag_set);
- if (rv) {
- tag_set.ops = NULL;
- goto out_cleanup_queues;
- }
+ if (dev->shared_tags) {
+ if (!tag_set.ops) {
+ rv = null_init_tag_set(NULL, &tag_set);
+ if (rv) {
+ tag_set.ops = NULL;
+ goto out_cleanup_queues;
}
- nullb->tag_set = &tag_set;
- rv = 0;
- } else {
- nullb->tag_set = &nullb->__tag_set;
- rv = null_init_tag_set(nullb, nullb->tag_set);
}
+ nullb->tag_set = &tag_set;
+ rv = 0;
+ } else {
+ nullb->tag_set = &nullb->__tag_set;
+ rv = null_init_tag_set(nullb, nullb->tag_set);
+ }
- if (rv)
- goto out_cleanup_queues;
-
- nullb->tag_set->timeout = 5 * HZ;
- nullb->disk = blk_mq_alloc_disk(nullb->tag_set, NULL, nullb);
- if (IS_ERR(nullb->disk)) {
- rv = PTR_ERR(nullb->disk);
- goto out_cleanup_tags;
- }
- nullb->q = nullb->disk->queue;
- } else if (dev->queue_mode == NULL_Q_BIO) {
- nullb->disk = blk_alloc_disk(NULL, nullb->dev->home_node);
- if (IS_ERR(nullb->disk)) {
- rv = PTR_ERR(nullb->disk);
- goto out_cleanup_queues;
- }
+ if (rv)
+ goto out_cleanup_queues;
- nullb->q = nullb->disk->queue;
- rv = init_driver_queues(nullb);
- if (rv)
- goto out_cleanup_disk;
+ nullb->tag_set->timeout = 5 * HZ;
+ nullb->disk = blk_mq_alloc_disk(nullb->tag_set, NULL, nullb);
+ if (IS_ERR(nullb->disk)) {
+ rv = PTR_ERR(nullb->disk);
+ goto out_cleanup_tags;
}
+ nullb->q = nullb->disk->queue;
if (dev->mbps) {
set_bit(NULLB_DEV_FL_THROTTLED, &dev->flags);
@@ -2232,10 +1995,10 @@ static int null_add_dev(struct nullb_device *dev)
out_cleanup_disk:
put_disk(nullb->disk);
out_cleanup_tags:
- if (dev->queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
+ if (nullb->tag_set == &nullb->__tag_set)
blk_mq_free_tag_set(nullb->tag_set);
out_cleanup_queues:
- cleanup_queues(nullb);
+ kfree(nullb->queues);
out_free_nullb:
kfree(nullb);
dev->nullb = NULL;
@@ -2311,7 +2074,7 @@ static int __init null_init(void)
return -EINVAL;
}
- if (g_queue_mode == NULL_Q_MQ && g_use_per_node_hctx) {
+ if (g_use_per_node_hctx) {
if (g_submit_queues != nr_online_nodes) {
pr_warn("submit_queues param is set to %u.\n",
nr_online_nodes);
@@ -16,11 +16,6 @@
#include <linux/mutex.h>
struct nullb_cmd {
- union {
- struct request *rq;
- struct bio *bio;
- };
- unsigned int tag;
blk_status_t error;
bool fake_timeout;
struct nullb_queue *nq;
@@ -28,16 +23,11 @@ struct nullb_cmd {
};
struct nullb_queue {
- unsigned long *tag_map;
- wait_queue_head_t wait;
- unsigned int queue_depth;
struct nullb_device *dev;
unsigned int requeue_selection;
struct list_head poll_list;
spinlock_t poll_lock;
-
- struct nullb_cmd *cmds;
};
struct nullb_zone {
@@ -60,13 +50,6 @@ struct nullb_zone {
unsigned int capacity;
};
-/* Queue modes */
-enum {
- NULL_Q_BIO = 0,
- NULL_Q_RQ = 1,
- NULL_Q_MQ = 2,
-};
-
struct nullb_device {
struct nullb *nullb;
struct config_group group;
@@ -41,10 +41,11 @@ TRACE_EVENT(nullb_zone_op,
__field(unsigned int, zone_cond)
),
TP_fast_assign(
- __entry->op = req_op(cmd->rq);
+ __entry->op = req_op(blk_mq_rq_from_pdu(cmd));
__entry->zone_no = zone_no;
__entry->zone_cond = zone_cond;
- __assign_disk_name(__entry->disk, cmd->rq->q->disk);
+ __assign_disk_name(__entry->disk,
+ blk_mq_rq_from_pdu(cmd)->q->disk);
),
TP_printk("%s req=%-15s zone_no=%u zone_cond=%-10s",
__print_disk_name(__entry->disk),
@@ -168,10 +168,7 @@ int null_register_zoned_dev(struct nullb *nullb)
disk_set_max_open_zones(nullb->disk, dev->zone_max_open);
disk_set_max_active_zones(nullb->disk, dev->zone_max_active);
- if (queue_is_mq(q))
- return blk_revalidate_disk_zones(nullb->disk, NULL);
-
- return 0;
+ return blk_revalidate_disk_zones(nullb->disk, NULL);
}
void null_free_zoned_dev(struct nullb_device *dev)
@@ -394,10 +391,7 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
*/
if (append) {
sector = zone->wp;
- if (dev->queue_mode == NULL_Q_MQ)
- cmd->rq->__sector = sector;
- else
- cmd->bio->bi_iter.bi_sector = sector;
+ blk_mq_rq_from_pdu(cmd)->__sector = sector;
} else if (sector != zone->wp) {
ret = BLK_STS_IOERR;
goto unlock;