aoe: convert aoeblk to blk-mq

Straight forward conversion - instead of rewriting the internal buffer
retrieval logic, just replace the previous elevator peeking with an
internal list of requests.

Reviewed-by: "Ed L. Cashin" <ed.cashin@acm.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

4 files changed, 61 insertions, 25 deletions

diff --git a/drivers/block/aoe/aoe.h b/drivers/block/aoe/aoe.h
index 015c68017a1c..7ca76ed2e71a 100644
--- a/drivers/block/aoe/aoe.h
+++ b/drivers/block/aoe/aoe.h
@@ -1,4 +1,6 @@
 /* Copyright (c) 2013 Coraid, Inc.  See COPYING for GPL terms. */
+#include <linux/blk-mq.h>
+
 #define VERSION "85"
 #define AOE_MAJOR 152
 #define DEVICE_NAME "aoe"
@@ -164,6 +166,8 @@ struct aoedev {
         struct gendisk *gd;
         struct dentry *debugfs;
         struct request_queue *blkq;
+        struct list_head rq_list;
+        struct blk_mq_tag_set tag_set;
         struct hd_geometry geo;
         sector_t ssize;
         struct timer_list timer;
diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c
index ff770e7d9e52..ed26b7287256 100644
--- a/drivers/block/aoe/aoeblk.c
+++ b/drivers/block/aoe/aoeblk.c
@@ -6,7 +6,7 @@
 
 #include <linux/kernel.h>
 #include <linux/hdreg.h>
-#include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/backing-dev.h>
 #include <linux/fs.h>
 #include <linux/ioctl.h>
@@ -268,23 +268,25 @@ aoeblk_release(struct gendisk *disk, fmode_t mode)
         spin_unlock_irqrestore(&d->lock, flags);
 }
 
-static void
-aoeblk_request(struct request_queue *q)
+static blk_status_t aoeblk_queue_rq(struct blk_mq_hw_ctx *hctx,
+                                    const struct blk_mq_queue_data *bd)
 {
-        struct aoedev *d;
-        struct request *rq;
+        struct aoedev *d = hctx->queue->queuedata;
+
+        spin_lock_irq(&d->lock);
 
-        d = q->queuedata;
         if ((d->flags & DEVFL_UP) == 0) {
                 pr_info_ratelimited("aoe: device %ld.%d is not up\n",
                         d->aoemajor, d->aoeminor);
-                while ((rq = blk_peek_request(q))) {
-                        blk_start_request(rq);
-                        aoe_end_request(d, rq, 1);
-                }
-                return;
+                spin_unlock_irq(&d->lock);
+                blk_mq_start_request(bd->rq);
+                return BLK_STS_IOERR;
         }
+
+        list_add_tail(&bd->rq->queuelist, &d->rq_list);
         aoecmd_work(d);
+        spin_unlock_irq(&d->lock);
+        return BLK_STS_OK;
 }
 
 static int
@@ -339,6 +341,10 @@ static const struct block_device_operations aoe_bdops = {
         .owner = THIS_MODULE,
 };
 
+static const struct blk_mq_ops aoeblk_mq_ops = {
+        .queue_rq       = aoeblk_queue_rq,
+};
+
 /* alloc_disk and add_disk can sleep */
 void
 aoeblk_gdalloc(void *vp)
@@ -347,9 +353,11 @@ aoeblk_gdalloc(void *vp)
         struct gendisk *gd;
         mempool_t *mp;
         struct request_queue *q;
+        struct blk_mq_tag_set *set;
         enum { KB = 1024, MB = KB * KB, READ_AHEAD = 2 * MB, };
         ulong flags;
         int late = 0;
+        int err;
 
         spin_lock_irqsave(&d->lock, flags);
         if (d->flags & DEVFL_GDALLOC
@@ -376,10 +384,25 @@ aoeblk_gdalloc(void *vp)
                         d->aoemajor, d->aoeminor);
                 goto err_disk;
         }
-        q = blk_init_queue(aoeblk_request, &d->lock);
-        if (q == NULL) {
+
+        set = &d->tag_set;
+        set->ops = &aoeblk_mq_ops;
+        set->nr_hw_queues = 1;
+        set->queue_depth = 128;
+        set->numa_node = NUMA_NO_NODE;
+        set->flags = BLK_MQ_F_SHOULD_MERGE;
+        err = blk_mq_alloc_tag_set(set);
+        if (err) {
+                pr_err("aoe: cannot allocate tag set for %ld.%d\n",
+                        d->aoemajor, d->aoeminor);
+                goto err_mempool;
+        }
+
+        q = blk_mq_init_queue(set);
+        if (IS_ERR(q)) {
                 pr_err("aoe: cannot allocate block queue for %ld.%d\n",
                         d->aoemajor, d->aoeminor);
+                blk_mq_free_tag_set(set);
                 goto err_mempool;
         }
 
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
index 136dc507d020..bb2fba651bd2 100644
--- a/drivers/block/aoe/aoecmd.c
+++ b/drivers/block/aoe/aoecmd.c
@@ -7,7 +7,7 @@
 #include <linux/ata.h>
 #include <linux/slab.h>
 #include <linux/hdreg.h>
-#include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/skbuff.h>
 #include <linux/netdevice.h>
 #include <linux/genhd.h>
@@ -813,7 +813,7 @@ rexmit_timer(struct timer_list *timer)
 out:
         if ((d->flags & DEVFL_KICKME) && d->blkq) {
                 d->flags &= ~DEVFL_KICKME;
-                d->blkq->request_fn(d->blkq);
+                blk_mq_run_hw_queues(d->blkq, true);
         }
 
         d->timer.expires = jiffies + TIMERTICK;
@@ -857,10 +857,12 @@ nextbuf(struct aoedev *d)
                 return d->ip.buf;
         rq = d->ip.rq;
         if (rq == NULL) {
-                rq = blk_peek_request(q);
+                rq = list_first_entry_or_null(&d->rq_list, struct request,
+                                                queuelist);
                 if (rq == NULL)
                         return NULL;
-                blk_start_request(rq);
+                list_del_init(&rq->queuelist);
+                blk_mq_start_request(rq);
                 d->ip.rq = rq;
                 d->ip.nxbio = rq->bio;
                 rq->special = (void *) rqbiocnt(rq);
@@ -1045,6 +1047,7 @@ aoe_end_request(struct aoedev *d, struct request *rq, int fastfail)
         struct bio *bio;
         int bok;
         struct request_queue *q;
+        blk_status_t err = BLK_STS_OK;
 
         q = d->blkq;
         if (rq == d->ip.rq)
@@ -1052,11 +1055,15 @@ aoe_end_request(struct aoedev *d, struct request *rq, int fastfail)
         do {
                 bio = rq->bio;
                 bok = !fastfail && !bio->bi_status;
-        } while (__blk_end_request(rq, bok ? BLK_STS_OK : BLK_STS_IOERR, bio->bi_iter.bi_size));
+                if (!bok)
+                        err = BLK_STS_IOERR;
+        } while (blk_update_request(rq, bok ? BLK_STS_OK : BLK_STS_IOERR, bio->bi_iter.bi_size));
+
+        __blk_mq_end_request(rq, err);
 
         /* cf. http://lkml.org/lkml/2006/10/31/28 */
         if (!fastfail)
-                __blk_run_queue(q);
+                blk_mq_run_hw_queues(q, true);
 }
 
 static void
diff --git a/drivers/block/aoe/aoedev.c b/drivers/block/aoe/aoedev.c
index f29a140cdbc1..9063f8efbd3b 100644
--- a/drivers/block/aoe/aoedev.c
+++ b/drivers/block/aoe/aoedev.c
@@ -5,7 +5,7 @@
  */
 
 #include <linux/hdreg.h>
-#include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/netdevice.h>
 #include <linux/delay.h>
 #include <linux/slab.h>
@@ -197,7 +197,6 @@ aoedev_downdev(struct aoedev *d)
 {
         struct aoetgt *t, **tt, **te;
         struct list_head *head, *pos, *nx;
-        struct request *rq;
         int i;
 
         d->flags &= ~DEVFL_UP;
@@ -225,10 +224,11 @@ aoedev_downdev(struct aoedev *d)
 
         /* fast fail all pending I/O */
         if (d->blkq) {
-                while ((rq = blk_peek_request(d->blkq))) {
-                        blk_start_request(rq);
-                        aoe_end_request(d, rq, 1);
-                }
+                /* UP is cleared, freeze+quiesce to insure all are errored */
+                blk_mq_freeze_queue(d->blkq);
+                blk_mq_quiesce_queue(d->blkq);
+                blk_mq_unquiesce_queue(d->blkq);
+                blk_mq_unfreeze_queue(d->blkq);
         }
 
         if (d->gd)
@@ -277,6 +277,7 @@ freedev(struct aoedev *d)
                 aoedisk_rm_debugfs(d);
                 del_gendisk(d->gd);
                 put_disk(d->gd);
+                blk_mq_free_tag_set(&d->tag_set);
                 blk_cleanup_queue(d->blkq);
         }
         t = d->targets;
@@ -463,6 +464,7 @@ aoedev_by_aoeaddr(ulong maj, int min, int do_alloc)
         d->ntargets = NTARGETS;
         INIT_WORK(&d->work, aoecmd_sleepwork);
         spin_lock_init(&d->lock);
+        INIT_LIST_HEAD(&d->rq_list);
         skb_queue_head_init(&d->skbpool);
         timer_setup(&d->timer, dummy_timer, 0);
         d->timer.expires = jiffies + HZ;