From: Yu Kuai <[email protected]>
Changes in v2:
- don't use test/set_bit() for new field, because unsigned long will
cause that some field can't be placed in the first cacheline(64 bytes),
use unsigned int for new field and test/set/clear it like 'bio->bi_flags'.
Yu Kuai (2):
block: introduce new field flags in block_device
block: warn once for each partition in bio_check_ro()
block/bdev.c | 15 ++++++++-------
block/blk-core.c | 21 +++++++++++++++------
block/genhd.c | 15 +++++++++++----
block/ioctl.c | 6 +++++-
include/linux/blk_types.h | 28 ++++++++++++++++++++++------
include/linux/blkdev.h | 5 +++--
6 files changed, 64 insertions(+), 26 deletions(-)
--
2.39.2
From: Yu Kuai <[email protected]>
Commit 1b0a151c10a6 ("blk-core: use pr_warn_ratelimited() in
bio_check_ro()") fix message storm by limit the rate, however, there
will still be lots of message in the long term. Fix it better by warn
once for each partition.
Signed-off-by: Yu Kuai <[email protected]>
---
block/blk-core.c | 14 +++++++++++---
include/linux/blk_types.h | 1 +
2 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/block/blk-core.c b/block/blk-core.c
index f9f8b12ba626..6575f684d41e 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -502,9 +502,17 @@ static inline void bio_check_ro(struct bio *bio)
if (op_is_write(bio_op(bio)) && bdev_read_only(bio->bi_bdev)) {
if (op_is_flush(bio->bi_opf) && !bio_sectors(bio))
return;
- pr_warn_ratelimited("Trying to write to read-only block-device %pg\n",
- bio->bi_bdev);
- /* Older lvm-tools actually trigger this */
+
+ if (bdev_flagged(bio->bi_bdev, BD_FLAG_RO_WARNED))
+ return;
+
+ bdev_set_flag(bio->bi_bdev, BD_FLAG_RO_WARNED);
+ /*
+ * Use ioctl to set underlying disk of raid/dm to read-only
+ * will trigger this.
+ */
+ pr_warn("Trying to write to read-only block-device %pg\n",
+ bio->bi_bdev);
}
}
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index f7a7965eeb25..07abd0165226 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -41,6 +41,7 @@ struct bio_crypt_ctx;
#define BD_FLAG_WRITE_HOLDER 1
#define BD_FLAG_HAS_SUBMIT_BIO 2
#define BD_FLAG_MAKE_IT_FAIL 3
+#define BD_FLAG_RO_WARNED 4
struct block_device {
sector_t bd_start_sect;
--
2.39.2
From: Yu Kuai <[email protected]>
There are multiple switches in struct block_device, use separate bool
fields for them is not gracefully. Add a new field flags and replace
swithes to a bit, there are no functional changes, and preare to add
a new switch in the next patch.
In order to keep flags in the first cacheline, define it as u32 and add
new helpers to set/clear/test each bit like 'bio->bi_flags'.
Signed-off-by: Yu Kuai <[email protected]>
---
block/bdev.c | 15 ++++++++-------
block/blk-core.c | 7 ++++---
block/genhd.c | 15 +++++++++++----
block/ioctl.c | 6 +++++-
include/linux/blk_types.h | 27 +++++++++++++++++++++------
include/linux/blkdev.h | 5 +++--
6 files changed, 52 insertions(+), 23 deletions(-)
diff --git a/block/bdev.c b/block/bdev.c
index fc8d28d77495..6b420cbd4f3d 100644
--- a/block/bdev.c
+++ b/block/bdev.c
@@ -408,10 +408,10 @@ struct block_device *bdev_alloc(struct gendisk *disk, u8 partno)
bdev->bd_partno = partno;
bdev->bd_inode = inode;
bdev->bd_queue = disk->queue;
- if (partno)
- bdev->bd_has_submit_bio = disk->part0->bd_has_submit_bio;
+ if (partno && bdev_flagged(disk->part0, BD_FLAG_HAS_SUBMIT_BIO))
+ bdev_set_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
else
- bdev->bd_has_submit_bio = false;
+ bdev_clear_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
bdev->bd_stats = alloc_percpu(struct disk_stats);
if (!bdev->bd_stats) {
iput(inode);
@@ -612,7 +612,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
bdev->bd_holder = NULL;
bdev->bd_holder_ops = NULL;
mutex_unlock(&bdev->bd_holder_lock);
- if (bdev->bd_write_holder)
+ if (bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER))
unblock = true;
}
if (!whole->bd_holders)
@@ -625,7 +625,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
*/
if (unblock) {
disk_unblock_events(bdev->bd_disk);
- bdev->bd_write_holder = false;
+ bdev_clear_flag(bdev, BD_FLAG_WRITE_HOLDER);
}
}
@@ -878,9 +878,10 @@ struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
* writeable reference is too fragile given the way @mode is
* used in blkdev_get/put().
*/
- if ((mode & BLK_OPEN_WRITE) && !bdev->bd_write_holder &&
+ if ((mode & BLK_OPEN_WRITE) &&
+ !bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER) &&
(disk->event_flags & DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE)) {
- bdev->bd_write_holder = true;
+ bdev_set_flag(bdev, BD_FLAG_WRITE_HOLDER);
unblock_events = false;
}
}
diff --git a/block/blk-core.c b/block/blk-core.c
index fdf25b8d6e78..f9f8b12ba626 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -482,7 +482,8 @@ __setup("fail_make_request=", setup_fail_make_request);
bool should_fail_request(struct block_device *part, unsigned int bytes)
{
- return part->bd_make_it_fail && should_fail(&fail_make_request, bytes);
+ return bdev_flagged(part, BD_FLAG_MAKE_IT_FAIL) &&
+ should_fail(&fail_make_request, bytes);
}
static int __init fail_make_request_debugfs(void)
@@ -595,7 +596,7 @@ static void __submit_bio(struct bio *bio)
if (unlikely(!blk_crypto_bio_prep(&bio)))
return;
- if (!bio->bi_bdev->bd_has_submit_bio) {
+ if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO)) {
blk_mq_submit_bio(bio);
} else if (likely(bio_queue_enter(bio) == 0)) {
struct gendisk *disk = bio->bi_bdev->bd_disk;
@@ -703,7 +704,7 @@ void submit_bio_noacct_nocheck(struct bio *bio)
*/
if (current->bio_list)
bio_list_add(¤t->bio_list[0], bio);
- else if (!bio->bi_bdev->bd_has_submit_bio)
+ else if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO))
__submit_bio_noacct_mq(bio);
else
__submit_bio_noacct(bio);
diff --git a/block/genhd.c b/block/genhd.c
index c9d06f72c587..57f96c0c8da0 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -413,7 +413,8 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
elevator_init_mq(disk->queue);
/* Mark bdev as having a submit_bio, if needed */
- disk->part0->bd_has_submit_bio = disk->fops->submit_bio != NULL;
+ if (disk->fops->submit_bio)
+ bdev_set_flag(disk->part0, BD_FLAG_HAS_SUBMIT_BIO);
/*
* If the driver provides an explicit major number it also must provide
@@ -1062,7 +1063,8 @@ static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL);
ssize_t part_fail_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
- return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail);
+ return sprintf(buf, "%d\n",
+ bdev_flagged(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL));
}
ssize_t part_fail_store(struct device *dev,
@@ -1071,8 +1073,13 @@ ssize_t part_fail_store(struct device *dev,
{
int i;
- if (count > 0 && sscanf(buf, "%d", &i) > 0)
- dev_to_bdev(dev)->bd_make_it_fail = i;
+ if (count > 0 && sscanf(buf, "%d", &i) > 0) {
+ if (!i)
+ bdev_clear_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
+ else
+ bdev_set_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
+
+ }
return count;
}
diff --git a/block/ioctl.c b/block/ioctl.c
index 4160f4e6bd5b..a64440f4c96b 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -394,7 +394,11 @@ static int blkdev_roset(struct block_device *bdev, unsigned cmd,
if (ret)
return ret;
}
- bdev->bd_read_only = n;
+
+ if (!n)
+ bdev_clear_flag(bdev, BD_FLAG_READ_ONLY);
+ else
+ bdev_set_flag(bdev, BD_FLAG_READ_ONLY);
return 0;
}
diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
index 52e264d5a830..f7a7965eeb25 100644
--- a/include/linux/blk_types.h
+++ b/include/linux/blk_types.h
@@ -37,6 +37,11 @@ struct bio_crypt_ctx;
#define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
#define SECTOR_MASK (PAGE_SECTORS - 1)
+#define BD_FLAG_READ_ONLY 0 /* read-only-policy */
+#define BD_FLAG_WRITE_HOLDER 1
+#define BD_FLAG_HAS_SUBMIT_BIO 2
+#define BD_FLAG_MAKE_IT_FAIL 3
+
struct block_device {
sector_t bd_start_sect;
sector_t bd_nr_sectors;
@@ -44,10 +49,8 @@ struct block_device {
struct request_queue * bd_queue;
struct disk_stats __percpu *bd_stats;
unsigned long bd_stamp;
- bool bd_read_only; /* read-only policy */
+ unsigned int bd_flags;
u8 bd_partno;
- bool bd_write_holder;
- bool bd_has_submit_bio;
dev_t bd_dev;
atomic_t bd_openers;
spinlock_t bd_size_lock; /* for bd_inode->i_size updates */
@@ -63,9 +66,6 @@ struct block_device {
struct mutex bd_fsfreeze_mutex; /* serialize freeze/thaw */
struct partition_meta_info *bd_meta_info;
-#ifdef CONFIG_FAIL_MAKE_REQUEST
- bool bd_make_it_fail;
-#endif
int bd_writers;
/*
* keep this out-of-line as it's both big and not needed in the fast
@@ -83,6 +83,21 @@ struct block_device {
#define bdev_kobj(_bdev) \
(&((_bdev)->bd_device.kobj))
+static inline bool bdev_flagged(struct block_device *bdev, unsigned int bit)
+{
+ return bdev->bd_flags & (1U << bit);
+}
+
+static inline void bdev_set_flag(struct block_device *bdev, unsigned int bit)
+{
+ bdev->bd_flags |= (1U << bit);
+}
+
+static inline void bdev_clear_flag(struct block_device *bdev, unsigned int bit)
+{
+ bdev->bd_flags &= ~(1U << bit);
+}
+
/*
* Block error status values. See block/blk-core:blk_errors for the details.
* Alpha cannot write a byte atomically, so we need to use 32-bit value.
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 9f6c3373f9fc..be1a39c0f8db 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -743,13 +743,14 @@ void disk_uevent(struct gendisk *disk, enum kobject_action action);
static inline int get_disk_ro(struct gendisk *disk)
{
- return disk->part0->bd_read_only ||
+ return bdev_flagged(disk->part0, BD_FLAG_READ_ONLY) ||
test_bit(GD_READ_ONLY, &disk->state);
}
static inline int bdev_read_only(struct block_device *bdev)
{
- return bdev->bd_read_only || get_disk_ro(bdev->bd_disk);
+ return bdev_flagged(bdev, BD_FLAG_READ_ONLY) ||
+ get_disk_ro(bdev->bd_disk);
}
bool set_capacity_and_notify(struct gendisk *disk, sector_t size);
--
2.39.2
On Tue, Nov 21, 2023 at 07:19:26PM +0800, Yu Kuai wrote:
> From: Yu Kuai <[email protected]>
>
> There are multiple switches in struct block_device, use separate bool
> fields for them is not gracefully. Add a new field flags and replace
> swithes to a bit, there are no functional changes, and preare to add
> a new switch in the next patch.
>
> In order to keep flags in the first cacheline, define it as u32 and add
> new helpers to set/clear/test each bit like 'bio->bi_flags'.
>
> Signed-off-by: Yu Kuai <[email protected]>
> ---
> block/bdev.c | 15 ++++++++-------
> block/blk-core.c | 7 ++++---
> block/genhd.c | 15 +++++++++++----
> block/ioctl.c | 6 +++++-
> include/linux/blk_types.h | 27 +++++++++++++++++++++------
> include/linux/blkdev.h | 5 +++--
> 6 files changed, 52 insertions(+), 23 deletions(-)
>
> diff --git a/block/bdev.c b/block/bdev.c
> index fc8d28d77495..6b420cbd4f3d 100644
> --- a/block/bdev.c
> +++ b/block/bdev.c
> @@ -408,10 +408,10 @@ struct block_device *bdev_alloc(struct gendisk *disk, u8 partno)
> bdev->bd_partno = partno;
> bdev->bd_inode = inode;
> bdev->bd_queue = disk->queue;
> - if (partno)
> - bdev->bd_has_submit_bio = disk->part0->bd_has_submit_bio;
> + if (partno && bdev_flagged(disk->part0, BD_FLAG_HAS_SUBMIT_BIO))
> + bdev_set_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
> else
> - bdev->bd_has_submit_bio = false;
> + bdev_clear_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
> bdev->bd_stats = alloc_percpu(struct disk_stats);
> if (!bdev->bd_stats) {
> iput(inode);
> @@ -612,7 +612,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
> bdev->bd_holder = NULL;
> bdev->bd_holder_ops = NULL;
> mutex_unlock(&bdev->bd_holder_lock);
> - if (bdev->bd_write_holder)
> + if (bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER))
> unblock = true;
> }
> if (!whole->bd_holders)
> @@ -625,7 +625,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
> */
> if (unblock) {
> disk_unblock_events(bdev->bd_disk);
> - bdev->bd_write_holder = false;
> + bdev_clear_flag(bdev, BD_FLAG_WRITE_HOLDER);
> }
> }
>
> @@ -878,9 +878,10 @@ struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
> * writeable reference is too fragile given the way @mode is
> * used in blkdev_get/put().
> */
> - if ((mode & BLK_OPEN_WRITE) && !bdev->bd_write_holder &&
> + if ((mode & BLK_OPEN_WRITE) &&
> + !bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER) &&
> (disk->event_flags & DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE)) {
> - bdev->bd_write_holder = true;
> + bdev_set_flag(bdev, BD_FLAG_WRITE_HOLDER);
> unblock_events = false;
> }
> }
> diff --git a/block/blk-core.c b/block/blk-core.c
> index fdf25b8d6e78..f9f8b12ba626 100644
> --- a/block/blk-core.c
> +++ b/block/blk-core.c
> @@ -482,7 +482,8 @@ __setup("fail_make_request=", setup_fail_make_request);
>
> bool should_fail_request(struct block_device *part, unsigned int bytes)
> {
> - return part->bd_make_it_fail && should_fail(&fail_make_request, bytes);
> + return bdev_flagged(part, BD_FLAG_MAKE_IT_FAIL) &&
> + should_fail(&fail_make_request, bytes);
> }
>
> static int __init fail_make_request_debugfs(void)
> @@ -595,7 +596,7 @@ static void __submit_bio(struct bio *bio)
> if (unlikely(!blk_crypto_bio_prep(&bio)))
> return;
>
> - if (!bio->bi_bdev->bd_has_submit_bio) {
> + if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO)) {
> blk_mq_submit_bio(bio);
> } else if (likely(bio_queue_enter(bio) == 0)) {
> struct gendisk *disk = bio->bi_bdev->bd_disk;
> @@ -703,7 +704,7 @@ void submit_bio_noacct_nocheck(struct bio *bio)
> */
> if (current->bio_list)
> bio_list_add(¤t->bio_list[0], bio);
> - else if (!bio->bi_bdev->bd_has_submit_bio)
> + else if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO))
> __submit_bio_noacct_mq(bio);
> else
> __submit_bio_noacct(bio);
> diff --git a/block/genhd.c b/block/genhd.c
> index c9d06f72c587..57f96c0c8da0 100644
> --- a/block/genhd.c
> +++ b/block/genhd.c
> @@ -413,7 +413,8 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
> elevator_init_mq(disk->queue);
>
> /* Mark bdev as having a submit_bio, if needed */
> - disk->part0->bd_has_submit_bio = disk->fops->submit_bio != NULL;
> + if (disk->fops->submit_bio)
> + bdev_set_flag(disk->part0, BD_FLAG_HAS_SUBMIT_BIO);
>
> /*
> * If the driver provides an explicit major number it also must provide
> @@ -1062,7 +1063,8 @@ static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL);
> ssize_t part_fail_show(struct device *dev,
> struct device_attribute *attr, char *buf)
> {
> - return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail);
> + return sprintf(buf, "%d\n",
> + bdev_flagged(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL));
> }
>
> ssize_t part_fail_store(struct device *dev,
> @@ -1071,8 +1073,13 @@ ssize_t part_fail_store(struct device *dev,
> {
> int i;
>
> - if (count > 0 && sscanf(buf, "%d", &i) > 0)
> - dev_to_bdev(dev)->bd_make_it_fail = i;
> + if (count > 0 && sscanf(buf, "%d", &i) > 0) {
> + if (!i)
> + bdev_clear_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
> + else
> + bdev_set_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
> +
> + }
>
> return count;
> }
> diff --git a/block/ioctl.c b/block/ioctl.c
> index 4160f4e6bd5b..a64440f4c96b 100644
> --- a/block/ioctl.c
> +++ b/block/ioctl.c
> @@ -394,7 +394,11 @@ static int blkdev_roset(struct block_device *bdev, unsigned cmd,
> if (ret)
> return ret;
> }
> - bdev->bd_read_only = n;
> +
> + if (!n)
> + bdev_clear_flag(bdev, BD_FLAG_READ_ONLY);
> + else
> + bdev_set_flag(bdev, BD_FLAG_READ_ONLY);
> return 0;
> }
>
> diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
> index 52e264d5a830..f7a7965eeb25 100644
> --- a/include/linux/blk_types.h
> +++ b/include/linux/blk_types.h
> @@ -37,6 +37,11 @@ struct bio_crypt_ctx;
> #define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
> #define SECTOR_MASK (PAGE_SECTORS - 1)
>
> +#define BD_FLAG_READ_ONLY 0 /* read-only-policy */
> +#define BD_FLAG_WRITE_HOLDER 1
> +#define BD_FLAG_HAS_SUBMIT_BIO 2
> +#define BD_FLAG_MAKE_IT_FAIL 3
> +
> struct block_device {
> sector_t bd_start_sect;
> sector_t bd_nr_sectors;
> @@ -44,10 +49,8 @@ struct block_device {
> struct request_queue * bd_queue;
> struct disk_stats __percpu *bd_stats;
> unsigned long bd_stamp;
> - bool bd_read_only; /* read-only policy */
> + unsigned int bd_flags;
> u8 bd_partno;
> - bool bd_write_holder;
> - bool bd_has_submit_bio;
> dev_t bd_dev;
> atomic_t bd_openers;
> spinlock_t bd_size_lock; /* for bd_inode->i_size updates */
As I mentioned, you can convert the above/original 4 fields into single
'unsigned int', otherwise one extra 'int' filed is added. Then, 'bd_size_lock'
is moved to the 2nd cache line, which isn't big deal since the field isn't
touched in fast path. Anyway, it can be one future cleanup, so:
Reviewed-by: Ming Lei <[email protected]>
Thanks,
Ming
Hi,
?? 2023/11/21 17:33, Ming Lei ะด??:
> On Tue, Nov 21, 2023 at 07:19:26PM +0800, Yu Kuai wrote:
>> From: Yu Kuai <[email protected]>
>>
>> There are multiple switches in struct block_device, use separate bool
>> fields for them is not gracefully. Add a new field flags and replace
>> swithes to a bit, there are no functional changes, and preare to add
>> a new switch in the next patch.
>>
>> In order to keep flags in the first cacheline, define it as u32 and add
>> new helpers to set/clear/test each bit like 'bio->bi_flags'.
>>
>> Signed-off-by: Yu Kuai <[email protected]>
>> ---
>> block/bdev.c | 15 ++++++++-------
>> block/blk-core.c | 7 ++++---
>> block/genhd.c | 15 +++++++++++----
>> block/ioctl.c | 6 +++++-
>> include/linux/blk_types.h | 27 +++++++++++++++++++++------
>> include/linux/blkdev.h | 5 +++--
>> 6 files changed, 52 insertions(+), 23 deletions(-)
>>
>> diff --git a/block/bdev.c b/block/bdev.c
>> index fc8d28d77495..6b420cbd4f3d 100644
>> --- a/block/bdev.c
>> +++ b/block/bdev.c
>> @@ -408,10 +408,10 @@ struct block_device *bdev_alloc(struct gendisk *disk, u8 partno)
>> bdev->bd_partno = partno;
>> bdev->bd_inode = inode;
>> bdev->bd_queue = disk->queue;
>> - if (partno)
>> - bdev->bd_has_submit_bio = disk->part0->bd_has_submit_bio;
>> + if (partno && bdev_flagged(disk->part0, BD_FLAG_HAS_SUBMIT_BIO))
>> + bdev_set_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
>> else
>> - bdev->bd_has_submit_bio = false;
>> + bdev_clear_flag(bdev, BD_FLAG_HAS_SUBMIT_BIO);
>> bdev->bd_stats = alloc_percpu(struct disk_stats);
>> if (!bdev->bd_stats) {
>> iput(inode);
>> @@ -612,7 +612,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
>> bdev->bd_holder = NULL;
>> bdev->bd_holder_ops = NULL;
>> mutex_unlock(&bdev->bd_holder_lock);
>> - if (bdev->bd_write_holder)
>> + if (bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER))
>> unblock = true;
>> }
>> if (!whole->bd_holders)
>> @@ -625,7 +625,7 @@ static void bd_end_claim(struct block_device *bdev, void *holder)
>> */
>> if (unblock) {
>> disk_unblock_events(bdev->bd_disk);
>> - bdev->bd_write_holder = false;
>> + bdev_clear_flag(bdev, BD_FLAG_WRITE_HOLDER);
>> }
>> }
>>
>> @@ -878,9 +878,10 @@ struct bdev_handle *bdev_open_by_dev(dev_t dev, blk_mode_t mode, void *holder,
>> * writeable reference is too fragile given the way @mode is
>> * used in blkdev_get/put().
>> */
>> - if ((mode & BLK_OPEN_WRITE) && !bdev->bd_write_holder &&
>> + if ((mode & BLK_OPEN_WRITE) &&
>> + !bdev_flagged(bdev, BD_FLAG_WRITE_HOLDER) &&
>> (disk->event_flags & DISK_EVENT_FLAG_BLOCK_ON_EXCL_WRITE)) {
>> - bdev->bd_write_holder = true;
>> + bdev_set_flag(bdev, BD_FLAG_WRITE_HOLDER);
>> unblock_events = false;
>> }
>> }
>> diff --git a/block/blk-core.c b/block/blk-core.c
>> index fdf25b8d6e78..f9f8b12ba626 100644
>> --- a/block/blk-core.c
>> +++ b/block/blk-core.c
>> @@ -482,7 +482,8 @@ __setup("fail_make_request=", setup_fail_make_request);
>>
>> bool should_fail_request(struct block_device *part, unsigned int bytes)
>> {
>> - return part->bd_make_it_fail && should_fail(&fail_make_request, bytes);
>> + return bdev_flagged(part, BD_FLAG_MAKE_IT_FAIL) &&
>> + should_fail(&fail_make_request, bytes);
>> }
>>
>> static int __init fail_make_request_debugfs(void)
>> @@ -595,7 +596,7 @@ static void __submit_bio(struct bio *bio)
>> if (unlikely(!blk_crypto_bio_prep(&bio)))
>> return;
>>
>> - if (!bio->bi_bdev->bd_has_submit_bio) {
>> + if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO)) {
>> blk_mq_submit_bio(bio);
>> } else if (likely(bio_queue_enter(bio) == 0)) {
>> struct gendisk *disk = bio->bi_bdev->bd_disk;
>> @@ -703,7 +704,7 @@ void submit_bio_noacct_nocheck(struct bio *bio)
>> */
>> if (current->bio_list)
>> bio_list_add(¤t->bio_list[0], bio);
>> - else if (!bio->bi_bdev->bd_has_submit_bio)
>> + else if (!bdev_flagged(bio->bi_bdev, BD_FLAG_HAS_SUBMIT_BIO))
>> __submit_bio_noacct_mq(bio);
>> else
>> __submit_bio_noacct(bio);
>> diff --git a/block/genhd.c b/block/genhd.c
>> index c9d06f72c587..57f96c0c8da0 100644
>> --- a/block/genhd.c
>> +++ b/block/genhd.c
>> @@ -413,7 +413,8 @@ int __must_check device_add_disk(struct device *parent, struct gendisk *disk,
>> elevator_init_mq(disk->queue);
>>
>> /* Mark bdev as having a submit_bio, if needed */
>> - disk->part0->bd_has_submit_bio = disk->fops->submit_bio != NULL;
>> + if (disk->fops->submit_bio)
>> + bdev_set_flag(disk->part0, BD_FLAG_HAS_SUBMIT_BIO);
>>
>> /*
>> * If the driver provides an explicit major number it also must provide
>> @@ -1062,7 +1063,8 @@ static DEVICE_ATTR(diskseq, 0444, diskseq_show, NULL);
>> ssize_t part_fail_show(struct device *dev,
>> struct device_attribute *attr, char *buf)
>> {
>> - return sprintf(buf, "%d\n", dev_to_bdev(dev)->bd_make_it_fail);
>> + return sprintf(buf, "%d\n",
>> + bdev_flagged(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL));
>> }
>>
>> ssize_t part_fail_store(struct device *dev,
>> @@ -1071,8 +1073,13 @@ ssize_t part_fail_store(struct device *dev,
>> {
>> int i;
>>
>> - if (count > 0 && sscanf(buf, "%d", &i) > 0)
>> - dev_to_bdev(dev)->bd_make_it_fail = i;
>> + if (count > 0 && sscanf(buf, "%d", &i) > 0) {
>> + if (!i)
>> + bdev_clear_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
>> + else
>> + bdev_set_flag(dev_to_bdev(dev), BD_FLAG_MAKE_IT_FAIL);
>> +
>> + }
>>
>> return count;
>> }
>> diff --git a/block/ioctl.c b/block/ioctl.c
>> index 4160f4e6bd5b..a64440f4c96b 100644
>> --- a/block/ioctl.c
>> +++ b/block/ioctl.c
>> @@ -394,7 +394,11 @@ static int blkdev_roset(struct block_device *bdev, unsigned cmd,
>> if (ret)
>> return ret;
>> }
>> - bdev->bd_read_only = n;
>> +
>> + if (!n)
>> + bdev_clear_flag(bdev, BD_FLAG_READ_ONLY);
>> + else
>> + bdev_set_flag(bdev, BD_FLAG_READ_ONLY);
>> return 0;
>> }
>>
>> diff --git a/include/linux/blk_types.h b/include/linux/blk_types.h
>> index 52e264d5a830..f7a7965eeb25 100644
>> --- a/include/linux/blk_types.h
>> +++ b/include/linux/blk_types.h
>> @@ -37,6 +37,11 @@ struct bio_crypt_ctx;
>> #define PAGE_SECTORS (1 << PAGE_SECTORS_SHIFT)
>> #define SECTOR_MASK (PAGE_SECTORS - 1)
>>
>> +#define BD_FLAG_READ_ONLY 0 /* read-only-policy */
>> +#define BD_FLAG_WRITE_HOLDER 1
>> +#define BD_FLAG_HAS_SUBMIT_BIO 2
>> +#define BD_FLAG_MAKE_IT_FAIL 3
>> +
>> struct block_device {
>> sector_t bd_start_sect;
>> sector_t bd_nr_sectors;
>> @@ -44,10 +49,8 @@ struct block_device {
>> struct request_queue * bd_queue;
>> struct disk_stats __percpu *bd_stats;
>> unsigned long bd_stamp;
>> - bool bd_read_only; /* read-only policy */
>> + unsigned int bd_flags;
>> u8 bd_partno;
>> - bool bd_write_holder;
>> - bool bd_has_submit_bio;
>> dev_t bd_dev;
>> atomic_t bd_openers;
>> spinlock_t bd_size_lock; /* for bd_inode->i_size updates */
>
> As I mentioned, you can convert the above/original 4 fields into single
> 'unsigned int', otherwise one extra 'int' filed is added. Then, 'bd_size_lock'
> is moved to the 2nd cache line, which isn't big deal since the field isn't
> touched in fast path. Anyway, it can be one future cleanup, so:
Thanks for the explanation, if you don't mind, I can change bd_flags to
unsigned short to keep this behaviour in the next version.
Thanks,
Kuai
>
> Reviewed-by: Ming Lei <[email protected]>
>
> Thanks,
> Ming
>
>
> .
>