Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
Merge tag 'block-5.9-2020-08-14' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"A few fixes on the block side of things:
- Discard granularity fix (Coly)
- rnbd cleanups (Guoqing)
- md error handling fix (Dan)
- md sysfs fix (Junxiao)
- Fix flush request accounting, which caused an IO slowdown for some
configurations (Ming)
- Properly propagate loop flag for partition scanning (Lennart)"
* tag 'block-5.9-2020-08-14' of git://git.kernel.dk/linux-block:
block: fix double account of flush request's driver tag
loop: unset GENHD_FL_NO_PART_SCAN on LOOP_CONFIGURE
rnbd: no need to set bi_end_io in rnbd_bio_map_kern
rnbd: remove rnbd_dev_submit_io
md-cluster: Fix potential error pointer dereference in resize_bitmaps()
block: check queue's limits.discard_granularity in __blkdev_issue_discard()
md: get sysfs entry after redundancy attr group create
+62
-66
+9
-2
block/blk-flush.c
+9
-2
block/blk-flush.c
···
308
308
flush_rq->mq_ctx = first_rq->mq_ctx;
309
309
flush_rq->mq_hctx = first_rq->mq_hctx;
310
310
311
-
if (!q->elevator)
311
+
if (!q->elevator) {
312
312
flush_rq->tag = first_rq->tag;
313
-
else
313
+
314
+
/*
315
+
* We borrow data request's driver tag, so have to mark
316
+
* this flush request as INFLIGHT for avoiding double
317
+
* account of this driver tag
318
+
*/
319
+
flush_rq->rq_flags |= RQF_MQ_INFLIGHT;
320
+
} else
314
321
flush_rq->internal_tag = first_rq->internal_tag;
315
322
316
323
flush_rq->cmd_flags = REQ_OP_FLUSH | REQ_PREFLUSH;
+9
block/blk-lib.c
+9
block/blk-lib.c
···
47
47
op = REQ_OP_DISCARD;
48
48
}
49
49
50
+
/* In case the discard granularity isn't set by buggy device driver */
51
+
if (WARN_ON_ONCE(!q->limits.discard_granularity)) {
52
+
char dev_name[BDEVNAME_SIZE];
53
+
54
+
bdevname(bdev, dev_name);
55
+
pr_err_ratelimited("%s: Error: discard_granularity is 0.\n", dev_name);
56
+
return -EOPNOTSUPP;
57
+
}
58
+
50
59
bs_mask = (bdev_logical_block_size(bdev) >> 9) - 1;
51
60
if ((sector | nr_sects) & bs_mask)
52
61
return -EINVAL;
+2
drivers/block/loop.c
+2
drivers/block/loop.c
···
1171
1171
if (part_shift)
1172
1172
lo->lo_flags |= LO_FLAGS_PARTSCAN;
1173
1173
partscan = lo->lo_flags & LO_FLAGS_PARTSCAN;
1174
+
if (partscan)
1175
+
lo->lo_disk->flags &= ~GENHD_FL_NO_PART_SCAN;
1174
1176
1175
1177
/* Grab the block_device to prevent its destruction after we
1176
1178
* put /dev/loopXX inode. Later in __loop_clr_fd() we bdput(bdev).
+3
-34
drivers/block/rnbd/rnbd-srv-dev.c
+3
-34
drivers/block/rnbd/rnbd-srv-dev.c
···
45
45
kfree(dev);
46
46
}
47
47
48
-
static void rnbd_dev_bi_end_io(struct bio *bio)
48
+
void rnbd_dev_bi_end_io(struct bio *bio)
49
49
{
50
50
struct rnbd_dev_blk_io *io = bio->bi_private;
51
51
···
63
63
* Map the kernel address into a bio suitable for io to a block
64
64
* device. Returns an error pointer in case of error.
65
65
*/
66
-
static struct bio *rnbd_bio_map_kern(void *data, struct bio_set *bs,
67
-
unsigned int len, gfp_t gfp_mask)
66
+
struct bio *rnbd_bio_map_kern(void *data, struct bio_set *bs,
67
+
unsigned int len, gfp_t gfp_mask)
68
68
{
69
69
unsigned long kaddr = (unsigned long)data;
70
70
unsigned long end = (kaddr + len + PAGE_SIZE - 1) >> PAGE_SHIFT;
···
99
99
offset = 0;
100
100
}
101
101
102
-
bio->bi_end_io = bio_put;
103
102
return bio;
104
-
}
105
-
106
-
int rnbd_dev_submit_io(struct rnbd_dev *dev, sector_t sector, void *data,
107
-
size_t len, u32 bi_size, enum rnbd_io_flags flags,
108
-
short prio, void *priv)
109
-
{
110
-
struct rnbd_dev_blk_io *io;
111
-
struct bio *bio;
112
-
113
-
/* Generate bio with pages pointing to the rdma buffer */
114
-
bio = rnbd_bio_map_kern(data, dev->ibd_bio_set, len, GFP_KERNEL);
115
-
if (IS_ERR(bio))
116
-
return PTR_ERR(bio);
117
-
118
-
io = container_of(bio, struct rnbd_dev_blk_io, bio);
119
-
120
-
io->dev = dev;
121
-
io->priv = priv;
122
-
123
-
bio->bi_end_io = rnbd_dev_bi_end_io;
124
-
bio->bi_private = io;
125
-
bio->bi_opf = rnbd_to_bio_flags(flags);
126
-
bio->bi_iter.bi_sector = sector;
127
-
bio->bi_iter.bi_size = bi_size;
128
-
bio_set_prio(bio, prio);
129
-
bio_set_dev(bio, dev->bdev);
130
-
131
-
submit_bio(bio);
132
-
133
-
return 0;
134
103
}
+5
-14
drivers/block/rnbd/rnbd-srv-dev.h
+5
-14
drivers/block/rnbd/rnbd-srv-dev.h
···
41
41
42
42
void rnbd_endio(void *priv, int error);
43
43
44
+
void rnbd_dev_bi_end_io(struct bio *bio);
45
+
46
+
struct bio *rnbd_bio_map_kern(void *data, struct bio_set *bs,
47
+
unsigned int len, gfp_t gfp_mask);
48
+
44
49
static inline int rnbd_dev_get_max_segs(const struct rnbd_dev *dev)
45
50
{
46
51
return queue_max_segments(bdev_get_queue(dev->bdev));
···
79
74
{
80
75
return bdev_get_queue(dev->bdev)->limits.discard_alignment;
81
76
}
82
-
83
-
/**
84
-
* rnbd_dev_submit_io() - Submit an I/O to the disk
85
-
* @dev: device to that the I/O is submitted
86
-
* @sector: address to read/write data to
87
-
* @data: I/O data to write or buffer to read I/O date into
88
-
* @len: length of @data
89
-
* @bi_size: Amount of data that will be read/written
90
-
* @prio: IO priority
91
-
* @priv: private data passed to @io_fn
92
-
*/
93
-
int rnbd_dev_submit_io(struct rnbd_dev *dev, sector_t sector, void *data,
94
-
size_t len, u32 bi_size, enum rnbd_io_flags flags,
95
-
short prio, void *priv);
96
77
97
78
#endif /* RNBD_SRV_DEV_H */
+23
-9
drivers/block/rnbd/rnbd-srv.c
+23
-9
drivers/block/rnbd/rnbd-srv.c
···
124
124
struct rnbd_srv_sess_dev *sess_dev;
125
125
u32 dev_id;
126
126
int err;
127
+
struct rnbd_dev_blk_io *io;
128
+
struct bio *bio;
129
+
short prio;
127
130
128
131
priv = kmalloc(sizeof(*priv), GFP_KERNEL);
129
132
if (!priv)
···
145
142
priv->sess_dev = sess_dev;
146
143
priv->id = id;
147
144
148
-
err = rnbd_dev_submit_io(sess_dev->rnbd_dev, le64_to_cpu(msg->sector),
149
-
data, datalen, le32_to_cpu(msg->bi_size),
150
-
le32_to_cpu(msg->rw),
151
-
srv_sess->ver < RNBD_PROTO_VER_MAJOR ||
152
-
usrlen < sizeof(*msg) ?
153
-
0 : le16_to_cpu(msg->prio), priv);
154
-
if (unlikely(err)) {
155
-
rnbd_srv_err(sess_dev, "Submitting I/O to device failed, err: %d\n",
156
-
err);
145
+
/* Generate bio with pages pointing to the rdma buffer */
146
+
bio = rnbd_bio_map_kern(data, sess_dev->rnbd_dev->ibd_bio_set, datalen, GFP_KERNEL);
147
+
if (IS_ERR(bio)) {
148
+
rnbd_srv_err(sess_dev, "Failed to generate bio, err: %ld\n", PTR_ERR(bio));
157
149
goto sess_dev_put;
158
150
}
151
+
152
+
io = container_of(bio, struct rnbd_dev_blk_io, bio);
153
+
io->dev = sess_dev->rnbd_dev;
154
+
io->priv = priv;
155
+
156
+
bio->bi_end_io = rnbd_dev_bi_end_io;
157
+
bio->bi_private = io;
158
+
bio->bi_opf = rnbd_to_bio_flags(le32_to_cpu(msg->rw));
159
+
bio->bi_iter.bi_sector = le64_to_cpu(msg->sector);
160
+
bio->bi_iter.bi_size = le32_to_cpu(msg->bi_size);
161
+
prio = srv_sess->ver < RNBD_PROTO_VER_MAJOR ||
162
+
usrlen < sizeof(*msg) ? 0 : le16_to_cpu(msg->prio);
163
+
bio_set_prio(bio, prio);
164
+
bio_set_dev(bio, sess_dev->rnbd_dev->bdev);
165
+
166
+
submit_bio(bio);
159
167
160
168
return 0;
161
169
+1
drivers/md/md-cluster.c
+1
drivers/md/md-cluster.c
+10
-7
drivers/md/md.c
+10
-7
drivers/md/md.c
···
850
850
sysfs_remove_group(&mddev->kobj, &md_redundancy_group);
851
851
if (mddev->sysfs_action)
852
852
sysfs_put(mddev->sysfs_action);
853
+
if (mddev->sysfs_completed)
854
+
sysfs_put(mddev->sysfs_completed);
855
+
if (mddev->sysfs_degraded)
856
+
sysfs_put(mddev->sysfs_degraded);
853
857
mddev->sysfs_action = NULL;
858
+
mddev->sysfs_completed = NULL;
859
+
mddev->sysfs_degraded = NULL;
854
860
}
855
861
}
856
862
mddev->sysfs_active = 0;
···
4074
4068
pr_warn("md: cannot register extra attributes for %s\n",
4075
4069
mdname(mddev));
4076
4070
mddev->sysfs_action = sysfs_get_dirent(mddev->kobj.sd, "sync_action");
4071
+
mddev->sysfs_completed = sysfs_get_dirent_safe(mddev->kobj.sd, "sync_completed");
4072
+
mddev->sysfs_degraded = sysfs_get_dirent_safe(mddev->kobj.sd, "degraded");
4077
4073
}
4078
4074
if (oldpers->sync_request != NULL &&
4079
4075
pers->sync_request == NULL) {
···
5590
5582
5591
5583
if (mddev->sysfs_state)
5592
5584
sysfs_put(mddev->sysfs_state);
5593
-
if (mddev->sysfs_completed)
5594
-
sysfs_put(mddev->sysfs_completed);
5595
-
if (mddev->sysfs_degraded)
5596
-
sysfs_put(mddev->sysfs_degraded);
5597
5585
if (mddev->sysfs_level)
5598
5586
sysfs_put(mddev->sysfs_level);
5599
-
5600
5587
5601
5588
if (mddev->gendisk)
5602
5589
del_gendisk(mddev->gendisk);
···
5760
5757
if (!error && mddev->kobj.sd) {
5761
5758
kobject_uevent(&mddev->kobj, KOBJ_ADD);
5762
5759
mddev->sysfs_state = sysfs_get_dirent_safe(mddev->kobj.sd, "array_state");
5763
-
mddev->sysfs_completed = sysfs_get_dirent_safe(mddev->kobj.sd, "sync_completed");
5764
-
mddev->sysfs_degraded = sysfs_get_dirent_safe(mddev->kobj.sd, "degraded");
5765
5760
mddev->sysfs_level = sysfs_get_dirent_safe(mddev->kobj.sd, "level");
5766
5761
}
5767
5762
mddev_put(mddev);
···
6037
6036
pr_warn("md: cannot register extra attributes for %s\n",
6038
6037
mdname(mddev));
6039
6038
mddev->sysfs_action = sysfs_get_dirent_safe(mddev->kobj.sd, "sync_action");
6039
+
mddev->sysfs_completed = sysfs_get_dirent_safe(mddev->kobj.sd, "sync_completed");
6040
+
mddev->sysfs_degraded = sysfs_get_dirent_safe(mddev->kobj.sd, "degraded");
6040
6041
} else if (mddev->ro == 2) /* auto-readonly not meaningful */
6041
6042
mddev->ro = 0;
6042
6043