leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity
d01316d046
linux/drivers/md/dm-zoned-target.c

926 lines
22 KiB
C
Raw Normal View History

dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/*
* Copyright (C) 2017 Western Digital Corporation or its affiliates.
*
* This file is released under the GPL.
*/
#include "dm-zoned.h"
#include <linux/module.h>
#define DM_MSG_PREFIX "zoned"
#define DMZ_MIN_BIOS 8192
/*
* Zone BIO context.
*/
struct dmz_bioctx {
struct dmz_target *target;
struct dm_zone *zone;
struct bio *bio;
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_t ref;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
};
/*
* Chunk work descriptor.
*/
struct dm_chunk_work {
struct work_struct work;
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_t refcount;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
struct dmz_target *target;
unsigned int chunk;
struct bio_list bio_list;
};
/*
* Target descriptor.
*/
struct dmz_target {
struct dm_dev *ddev;
unsigned long flags;
/* Zoned block device information */
struct dmz_dev *dev;
/* For metadata handling */
struct dmz_metadata *metadata;
/* For reclaim */
struct dmz_reclaim *reclaim;
/* For chunk work */
struct radix_tree_root chunk_rxtree;
struct workqueue_struct *chunk_wq;
dm: adjust structure members to improve alignment Eliminate most holes in DM data structures that were modified by commit 6f1c819c21 ("dm: convert to bioset_init()/mempool_init()"). Also prevent structure members from unnecessarily spanning cache lines. Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-05-22 22:26:20 +00:00
struct mutex chunk_lock;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/* For cloned BIOs to zones */
dm: convert to bioset_init()/mempool_init() Convert dm to embedded bio sets. Acked-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-20 22:25:53 +00:00
struct bio_set bio_set;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/* For flush */
spinlock_t flush_lock;
struct bio_list flush_list;
struct delayed_work flush_work;
struct workqueue_struct *flush_wq;
};
/*
* Flush intervals (seconds).
*/
#define DMZ_FLUSH_PERIOD (10 * HZ)
/*
* Target BIO completion.
*/
static inline void dmz_bio_endio(struct bio *bio, blk_status_t status)
{
struct dmz_bioctx *bioctx = dm_per_bio_data(bio, sizeof(struct dmz_bioctx));
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
if (status != BLK_STS_OK && bio->bi_status == BLK_STS_OK)
bio->bi_status = status;
if (refcount_dec_and_test(&bioctx->ref)) {
struct dm_zone *zone = bioctx->zone;
if (zone) {
if (bio->bi_status != BLK_STS_OK &&
bio_op(bio) == REQ_OP_WRITE &&
dmz_is_seq(zone))
set_bit(DMZ_SEQ_WRITE_ERR, &zone->flags);
dmz_deactivate_zone(zone);
}
bio_endio(bio);
}
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
}
/*
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
* Completion callback for an internally cloned target BIO. This terminates the
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
* target BIO when there are no more references to its context.
*/
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
static void dmz_clone_endio(struct bio *clone)
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
{
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
struct dmz_bioctx *bioctx = clone->bi_private;
blk_status_t status = clone->bi_status;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
bio_put(clone);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dmz_bio_endio(bioctx->bio, status);
}
/*
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
* Issue a clone of a target BIO. The clone may only partially process the
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
* original target BIO.
*/
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
static int dmz_submit_bio(struct dmz_target *dmz, struct dm_zone *zone,
struct bio *bio, sector_t chunk_block,
unsigned int nr_blocks)
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
{
struct dmz_bioctx *bioctx = dm_per_bio_data(bio, sizeof(struct dmz_bioctx));
struct bio *clone;
dm: convert to bioset_init()/mempool_init() Convert dm to embedded bio sets. Acked-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-20 22:25:53 +00:00
clone = bio_clone_fast(bio, GFP_NOIO, &dmz->bio_set);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
if (!clone)
return -ENOMEM;
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
bio_set_dev(clone, dmz->dev->bdev);
clone->bi_iter.bi_sector =
dmz_start_sect(dmz->metadata, zone) + dmz_blk2sect(chunk_block);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
clone->bi_iter.bi_size = dmz_blk2sect(nr_blocks) << SECTOR_SHIFT;
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
clone->bi_end_io = dmz_clone_endio;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
clone->bi_private = bioctx;
bio_advance(bio, clone->bi_iter.bi_size);
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_inc(&bioctx->ref);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
generic_make_request(clone);
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
if (bio_op(bio) == REQ_OP_WRITE && dmz_is_seq(zone))
zone->wp_block += nr_blocks;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
return 0;
}
/*
* Zero out pages of discarded blocks accessed by a read BIO.
*/
static void dmz_handle_read_zero(struct dmz_target *dmz, struct bio *bio,
sector_t chunk_block, unsigned int nr_blocks)
{
unsigned int size = nr_blocks << DMZ_BLOCK_SHIFT;
/* Clear nr_blocks */
swap(bio->bi_iter.bi_size, size);
zero_fill_bio(bio);
swap(bio->bi_iter.bi_size, size);
bio_advance(bio, size);
}
/*
* Process a read BIO.
*/
static int dmz_handle_read(struct dmz_target *dmz, struct dm_zone *zone,
struct bio *bio)
{
sector_t chunk_block = dmz_chunk_block(dmz->dev, dmz_bio_block(bio));
unsigned int nr_blocks = dmz_bio_blocks(bio);
sector_t end_block = chunk_block + nr_blocks;
struct dm_zone *rzone, *bzone;
int ret;
/* Read into unmapped chunks need only zeroing the BIO buffer */
if (!zone) {
zero_fill_bio(bio);
return 0;
}
dmz_dev_debug(dmz->dev, "READ chunk %llu -> %s zone %u, block %llu, %u blocks",
(unsigned long long)dmz_bio_chunk(dmz->dev, bio),
(dmz_is_rnd(zone) ? "RND" : "SEQ"),
dmz_id(dmz->metadata, zone),
(unsigned long long)chunk_block, nr_blocks);
/* Check block validity to determine the read location */
bzone = zone->bzone;
while (chunk_block < end_block) {
nr_blocks = 0;
if (dmz_is_rnd(zone) || chunk_block < zone->wp_block) {
/* Test block validity in the data zone */
ret = dmz_block_valid(dmz->metadata, zone, chunk_block);
if (ret < 0)
return ret;
if (ret > 0) {
/* Read data zone blocks */
nr_blocks = ret;
rzone = zone;
}
}
/*
* No valid blocks found in the data zone.
* Check the buffer zone, if there is one.
*/
if (!nr_blocks && bzone) {
ret = dmz_block_valid(dmz->metadata, bzone, chunk_block);
if (ret < 0)
return ret;
if (ret > 0) {
/* Read buffer zone blocks */
nr_blocks = ret;
rzone = bzone;
}
}
if (nr_blocks) {
/* Valid blocks found: read them */
nr_blocks = min_t(unsigned int, nr_blocks, end_block - chunk_block);
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
ret = dmz_submit_bio(dmz, rzone, bio, chunk_block, nr_blocks);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
if (ret)
return ret;
chunk_block += nr_blocks;
} else {
/* No valid block: zeroout the current BIO block */
dmz_handle_read_zero(dmz, bio, chunk_block, 1);
chunk_block++;
}
}
return 0;
}
/*
* Write blocks directly in a data zone, at the write pointer.
* If a buffer zone is assigned, invalidate the blocks written
* in place.
*/
static int dmz_handle_direct_write(struct dmz_target *dmz,
struct dm_zone *zone, struct bio *bio,
sector_t chunk_block,
unsigned int nr_blocks)
{
struct dmz_metadata *zmd = dmz->metadata;
struct dm_zone *bzone = zone->bzone;
int ret;
if (dmz_is_readonly(zone))
return -EROFS;
/* Submit write */
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
ret = dmz_submit_bio(dmz, zone, bio, chunk_block, nr_blocks);
if (ret)
return ret;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/*
* Validate the blocks in the data zone and invalidate
* in the buffer zone, if there is one.
*/
ret = dmz_validate_blocks(zmd, zone, chunk_block, nr_blocks);
if (ret == 0 && bzone)
ret = dmz_invalidate_blocks(zmd, bzone, chunk_block, nr_blocks);
return ret;
}
/*
* Write blocks in the buffer zone of @zone.
* If no buffer zone is assigned yet, get one.
* Called with @zone write locked.
*/
static int dmz_handle_buffered_write(struct dmz_target *dmz,
struct dm_zone *zone, struct bio *bio,
sector_t chunk_block,
unsigned int nr_blocks)
{
struct dmz_metadata *zmd = dmz->metadata;
struct dm_zone *bzone;
int ret;
/* Get the buffer zone. One will be allocated if needed */
bzone = dmz_get_chunk_buffer(zmd, zone);
if (!bzone)
return -ENOSPC;
if (dmz_is_readonly(bzone))
return -EROFS;
/* Submit write */
dm zoned: Fix target BIO completion handling struct bioctx includes the ref refcount_t to track the number of I/O fragments used to process a target BIO as well as ensure that the zone of the BIO is kept in the active state throughout the lifetime of the BIO. However, since decrementing of this reference count is done in the target .end_io method, the function bio_endio() must be called multiple times for read and write target BIOs, which causes problems with the value of the __bi_remaining struct bio field for chained BIOs (e.g. the clone BIO passed by dm core is large and splits into fragments by the block layer), resulting in incorrect values and inconsistencies with the BIO_CHAIN flag setting. This is turn triggers the BUG_ON() call: BUG_ON(atomic_read(&bio->__bi_remaining) <= 0); in bio_remaining_done() called from bio_endio(). Fix this ensuring that bio_endio() is called only once for any target BIO by always using internal clone BIOs for processing any read or write target BIO. This allows reference counting using the target BIO context counter to trigger the target BIO completion bio_endio() call once all data, metadata and other zone work triggered by the BIO complete. Overall, this simplifies the code too as the target .end_io becomes unnecessary and differences between read and write BIO issuing and completion processing disappear. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-11-30 06:31:48 +00:00
ret = dmz_submit_bio(dmz, bzone, bio, chunk_block, nr_blocks);
if (ret)
return ret;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/*
* Validate the blocks in the buffer zone
* and invalidate in the data zone.
*/
ret = dmz_validate_blocks(zmd, bzone, chunk_block, nr_blocks);
if (ret == 0 && chunk_block < zone->wp_block)
ret = dmz_invalidate_blocks(zmd, zone, chunk_block, nr_blocks);
return ret;
}
/*
* Process a write BIO.
*/
static int dmz_handle_write(struct dmz_target *dmz, struct dm_zone *zone,
struct bio *bio)
{
sector_t chunk_block = dmz_chunk_block(dmz->dev, dmz_bio_block(bio));
unsigned int nr_blocks = dmz_bio_blocks(bio);
if (!zone)
return -ENOSPC;
dmz_dev_debug(dmz->dev, "WRITE chunk %llu -> %s zone %u, block %llu, %u blocks",
(unsigned long long)dmz_bio_chunk(dmz->dev, bio),
(dmz_is_rnd(zone) ? "RND" : "SEQ"),
dmz_id(dmz->metadata, zone),
(unsigned long long)chunk_block, nr_blocks);
if (dmz_is_rnd(zone) || chunk_block == zone->wp_block) {
/*
* zone is a random zone or it is a sequential zone
* and the BIO is aligned to the zone write pointer:
* direct write the zone.
*/
return dmz_handle_direct_write(dmz, zone, bio, chunk_block, nr_blocks);
}
/*
* This is an unaligned write in a sequential zone:
* use buffered write.
*/
return dmz_handle_buffered_write(dmz, zone, bio, chunk_block, nr_blocks);
}
/*
* Process a discard BIO.
*/
static int dmz_handle_discard(struct dmz_target *dmz, struct dm_zone *zone,
struct bio *bio)
{
struct dmz_metadata *zmd = dmz->metadata;
sector_t block = dmz_bio_block(bio);
unsigned int nr_blocks = dmz_bio_blocks(bio);
sector_t chunk_block = dmz_chunk_block(dmz->dev, block);
int ret = 0;
/* For unmapped chunks, there is nothing to do */
if (!zone)
return 0;
if (dmz_is_readonly(zone))
return -EROFS;
dmz_dev_debug(dmz->dev, "DISCARD chunk %llu -> zone %u, block %llu, %u blocks",
(unsigned long long)dmz_bio_chunk(dmz->dev, bio),
dmz_id(zmd, zone),
(unsigned long long)chunk_block, nr_blocks);
/*
* Invalidate blocks in the data zone and its
* buffer zone if one is mapped.
*/
if (dmz_is_rnd(zone) || chunk_block < zone->wp_block)
ret = dmz_invalidate_blocks(zmd, zone, chunk_block, nr_blocks);
if (ret == 0 && zone->bzone)
ret = dmz_invalidate_blocks(zmd, zone->bzone,
chunk_block, nr_blocks);
return ret;
}
/*
* Process a BIO.
*/
static void dmz_handle_bio(struct dmz_target *dmz, struct dm_chunk_work *cw,
struct bio *bio)
{
struct dmz_bioctx *bioctx = dm_per_bio_data(bio, sizeof(struct dmz_bioctx));
struct dmz_metadata *zmd = dmz->metadata;
struct dm_zone *zone;
int ret;
/*
* Write may trigger a zone allocation. So make sure the
* allocation can succeed.
*/
if (bio_op(bio) == REQ_OP_WRITE)
dmz_schedule_reclaim(dmz->reclaim);
dmz_lock_metadata(zmd);
/*
* Get the data zone mapping the chunk. There may be no
* mapping for read and discard. If a mapping is obtained,
+ the zone returned will be set to active state.
*/
zone = dmz_get_chunk_mapping(zmd, dmz_bio_chunk(dmz->dev, bio),
bio_op(bio));
if (IS_ERR(zone)) {
ret = PTR_ERR(zone);
goto out;
}
/* Process the BIO */
if (zone) {
dmz_activate_zone(zone);
bioctx->zone = zone;
}
switch (bio_op(bio)) {
case REQ_OP_READ:
ret = dmz_handle_read(dmz, zone, bio);
break;
case REQ_OP_WRITE:
ret = dmz_handle_write(dmz, zone, bio);
break;
case REQ_OP_DISCARD:
case REQ_OP_WRITE_ZEROES:
ret = dmz_handle_discard(dmz, zone, bio);
break;
default:
dmz_dev_err(dmz->dev, "Unsupported BIO operation 0x%x",
bio_op(bio));
ret = -EIO;
}
/*
* Release the chunk mapping. This will check that the mapping
* is still valid, that is, that the zone used still has valid blocks.
*/
if (zone)
dmz_put_chunk_mapping(zmd, zone);
out:
dmz_bio_endio(bio, errno_to_blk_status(ret));
dmz_unlock_metadata(zmd);
}
/*
* Increment a chunk reference counter.
*/
static inline void dmz_get_chunk_work(struct dm_chunk_work *cw)
{
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_inc(&cw->refcount);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
}
/*
* Decrement a chunk work reference count and
* free it if it becomes 0.
*/
static void dmz_put_chunk_work(struct dm_chunk_work *cw)
{
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
if (refcount_dec_and_test(&cw->refcount)) {
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
WARN_ON(!bio_list_empty(&cw->bio_list));
radix_tree_delete(&cw->target->chunk_rxtree, cw->chunk);
kfree(cw);
}
}
/*
* Chunk BIO work function.
*/
static void dmz_chunk_work(struct work_struct *work)
{
struct dm_chunk_work *cw = container_of(work, struct dm_chunk_work, work);
struct dmz_target *dmz = cw->target;
struct bio *bio;
mutex_lock(&dmz->chunk_lock);
/* Process the chunk BIOs */
while ((bio = bio_list_pop(&cw->bio_list))) {
mutex_unlock(&dmz->chunk_lock);
dmz_handle_bio(dmz, cw, bio);
mutex_lock(&dmz->chunk_lock);
dmz_put_chunk_work(cw);
}
/* Queueing the work incremented the work refcount */
dmz_put_chunk_work(cw);
mutex_unlock(&dmz->chunk_lock);
}
/*
* Flush work.
*/
static void dmz_flush_work(struct work_struct *work)
{
struct dmz_target *dmz = container_of(work, struct dmz_target, flush_work.work);
struct bio *bio;
int ret;
/* Flush dirty metadata blocks */
ret = dmz_flush_metadata(dmz->metadata);
/* Process queued flush requests */
while (1) {
spin_lock(&dmz->flush_lock);
bio = bio_list_pop(&dmz->flush_list);
spin_unlock(&dmz->flush_lock);
if (!bio)
break;
dmz_bio_endio(bio, errno_to_blk_status(ret));
}
queue_delayed_work(dmz->flush_wq, &dmz->flush_work, DMZ_FLUSH_PERIOD);
}
/*
* Get a chunk work and start it to process a new BIO.
* If the BIO chunk has no work yet, create one.
*/
static void dmz_queue_chunk_work(struct dmz_target *dmz, struct bio *bio)
{
unsigned int chunk = dmz_bio_chunk(dmz->dev, bio);
struct dm_chunk_work *cw;
mutex_lock(&dmz->chunk_lock);
/* Get the BIO chunk work. If one is not active yet, create one */
cw = radix_tree_lookup(&dmz->chunk_rxtree, chunk);
if (!cw) {
int ret;
/* Create a new chunk work */
dm zoned: use GFP_NOIO in I/O path Use GFP_NOIO for memory allocations in the I/O path. Other memory allocations in the initialization path can use GFP_KERNEL. Reported-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-07-24 07:44:37 +00:00
cw = kmalloc(sizeof(struct dm_chunk_work), GFP_NOIO);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
if (!cw)
goto out;
INIT_WORK(&cw->work, dmz_chunk_work);
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_set(&cw->refcount, 0);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
cw->target = dmz;
cw->chunk = chunk;
bio_list_init(&cw->bio_list);
ret = radix_tree_insert(&dmz->chunk_rxtree, chunk, cw);
if (unlikely(ret)) {
kfree(cw);
cw = NULL;
goto out;
}
}
bio_list_add(&cw->bio_list, bio);
dmz_get_chunk_work(cw);
if (queue_work(dmz->chunk_wq, &cw->work))
dmz_get_chunk_work(cw);
out:
mutex_unlock(&dmz->chunk_lock);
}
/*
* Process a new BIO.
*/
static int dmz_map(struct dm_target *ti, struct bio *bio)
{
struct dmz_target *dmz = ti->private;
struct dmz_dev *dev = dmz->dev;
struct dmz_bioctx *bioctx = dm_per_bio_data(bio, sizeof(struct dmz_bioctx));
sector_t sector = bio->bi_iter.bi_sector;
unsigned int nr_sectors = bio_sectors(bio);
sector_t chunk_sector;
dmz_dev_debug(dev, "BIO op %d sector %llu + %u => chunk %llu, block %llu, %u blocks",
bio_op(bio), (unsigned long long)sector, nr_sectors,
(unsigned long long)dmz_bio_chunk(dmz->dev, bio),
(unsigned long long)dmz_chunk_block(dmz->dev, dmz_bio_block(bio)),
(unsigned int)dmz_bio_blocks(bio));
block: replace bi_bdev with a gendisk pointer and partitions index This way we don't need a block_device structure to submit I/O. The block_device has different life time rules from the gendisk and request_queue and is usually only available when the block device node is open. Other callers need to explicitly create one (e.g. the lightnvm passthrough code, or the new nvme multipathing code). For the actual I/O path all that we need is the gendisk, which exists once per block device. But given that the block layer also does partition remapping we additionally need a partition index, which is used for said remapping in generic_make_request. Note that all the block drivers generally want request_queue or sometimes the gendisk, so this removes a layer of indirection all over the stack. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-08-23 17:10:32 +00:00
bio_set_dev(bio, dev->bdev);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dm zoned: remove test for impossible REQ_OP_FLUSH conditions The value REQ_OP_FLUSH is only used by the block code for request-based devices. Remove the tests for REQ_OP_FLUSH from the bio-based dm-zoned-target. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-07-21 15:56:46 +00:00
if (!nr_sectors && bio_op(bio) != REQ_OP_WRITE)
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
return DM_MAPIO_REMAPPED;
/* The BIO should be block aligned */
if ((nr_sectors & DMZ_BLOCK_SECTORS_MASK) || (sector & DMZ_BLOCK_SECTORS_MASK))
return DM_MAPIO_KILL;
/* Initialize the BIO context */
bioctx->target = dmz;
bioctx->zone = NULL;
bioctx->bio = bio;
dm zoned: target: use refcount_t for dm zoned reference counters The API surrounding refcount_t should be used in place of atomic_t when variables are being used as reference counters. This API can prevent issues such as counter overflows and use-after-free conditions. Within the dm zoned target stack, the atomic_t API is used for bioctx->ref and cw->refcount. Change these to use refcount_t, avoiding the issues mentioned. Signed-off-by: John Pittman <jpittman@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Tested-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-08-23 17:35:57 +00:00
refcount_set(&bioctx->ref, 1);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
/* Set the BIO pending in the flush list */
dm zoned: remove test for impossible REQ_OP_FLUSH conditions The value REQ_OP_FLUSH is only used by the block code for request-based devices. Remove the tests for REQ_OP_FLUSH from the bio-based dm-zoned-target. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Reviewed-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-07-21 15:56:46 +00:00
if (!nr_sectors && bio_op(bio) == REQ_OP_WRITE) {
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
spin_lock(&dmz->flush_lock);
bio_list_add(&dmz->flush_list, bio);
spin_unlock(&dmz->flush_lock);
mod_delayed_work(dmz->flush_wq, &dmz->flush_work, 0);
return DM_MAPIO_SUBMITTED;
}
/* Split zone BIOs to fit entirely into a zone */
chunk_sector = sector & (dev->zone_nr_sectors - 1);
if (chunk_sector + nr_sectors > dev->zone_nr_sectors)
dm_accept_partial_bio(bio, dev->zone_nr_sectors - chunk_sector);
/* Now ready to handle this BIO */
dmz_reclaim_bio_acc(dmz->reclaim);
dmz_queue_chunk_work(dmz, bio);
return DM_MAPIO_SUBMITTED;
}
/*
* Get zoned device information.
*/
static int dmz_get_zoned_device(struct dm_target *ti, char *path)
{
struct dmz_target *dmz = ti->private;
struct request_queue *q;
struct dmz_dev *dev;
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
sector_t aligned_capacity;
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
int ret;
/* Get the target device */
ret = dm_get_device(ti, path, dm_table_get_mode(ti->table), &dmz->ddev);
if (ret) {
ti->error = "Get target device failed";
dmz->ddev = NULL;
return ret;
}
dev = kzalloc(sizeof(struct dmz_dev), GFP_KERNEL);
if (!dev) {
ret = -ENOMEM;
goto err;
}
dev->bdev = dmz->ddev->bdev;
(void)bdevname(dev->bdev, dev->name);
if (bdev_zoned_model(dev->bdev) == BLK_ZONED_NONE) {
ti->error = "Not a zoned block device";
ret = -EINVAL;
goto err;
}
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
q = bdev_get_queue(dev->bdev);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dev->capacity = i_size_read(dev->bdev->bd_inode) >> SECTOR_SHIFT;
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
aligned_capacity = dev->capacity & ~(blk_queue_zone_sectors(q) - 1);
if (ti->begin ||
((ti->len != dev->capacity) && (ti->len != aligned_capacity))) {
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
ti->error = "Partial mapping not supported";
ret = -EINVAL;
goto err;
}
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
dev->zone_nr_sectors = blk_queue_zone_sectors(q);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dev->zone_nr_sectors_shift = ilog2(dev->zone_nr_sectors);
dev->zone_nr_blocks = dmz_sect2blk(dev->zone_nr_sectors);
dev->zone_nr_blocks_shift = ilog2(dev->zone_nr_blocks);
block: Introduce blkdev_nr_zones() helper Introduce the blkdev_nr_zones() helper function to get the total number of zones of a zoned block device. This number is always 0 for a regular block device (q->limits.zoned == BLK_ZONED_NONE case). Replace hard-coded number of zones calculation in dmz_get_zoned_device() with a call to this helper. Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.com> Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-10-12 10:08:43 +00:00
dev->nr_zones = blkdev_nr_zones(dev->bdev);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dmz->dev = dev;
return 0;
err:
dm_put_device(ti, dmz->ddev);
kfree(dev);
return ret;
}
/*
* Cleanup zoned device information.
*/
static void dmz_put_zoned_device(struct dm_target *ti)
{
struct dmz_target *dmz = ti->private;
dm_put_device(ti, dmz->ddev);
kfree(dmz->dev);
dmz->dev = NULL;
}
/*
* Setup target.
*/
static int dmz_ctr(struct dm_target *ti, unsigned int argc, char **argv)
{
struct dmz_target *dmz;
struct dmz_dev *dev;
int ret;
/* Check arguments */
if (argc != 1) {
ti->error = "Invalid argument count";
return -EINVAL;
}
/* Allocate and initialize the target descriptor */
dmz = kzalloc(sizeof(struct dmz_target), GFP_KERNEL);
if (!dmz) {
ti->error = "Unable to allocate the zoned target descriptor";
return -ENOMEM;
}
ti->private = dmz;
/* Get the target zoned block device */
ret = dmz_get_zoned_device(ti, argv[0]);
if (ret) {
dmz->ddev = NULL;
goto err;
}
/* Initialize metadata */
dev = dmz->dev;
ret = dmz_ctr_metadata(dev, &dmz->metadata);
if (ret) {
ti->error = "Metadata initialization failed";
goto err_dev;
}
/* Set target (no write same support) */
ti->max_io_len = dev->zone_nr_sectors << 9;
ti->num_flush_bios = 1;
ti->num_discard_bios = 1;
ti->num_write_zeroes_bios = 1;
ti->per_io_data_size = sizeof(struct dmz_bioctx);
ti->flush_supported = true;
ti->discards_supported = true;
/* The exposed capacity is the number of chunks that can be mapped */
ti->len = (sector_t)dmz_nr_chunks(dmz->metadata) << dev->zone_nr_sectors_shift;
/* Zone BIO */
dm: convert to bioset_init()/mempool_init() Convert dm to embedded bio sets. Acked-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-20 22:25:53 +00:00
ret = bioset_init(&dmz->bio_set, DMZ_MIN_BIOS, 0, 0);
if (ret) {
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
ti->error = "Create BIO set failed";
goto err_meta;
}
/* Chunk BIO work */
mutex_init(&dmz->chunk_lock);
dm zoned: avoid triggering reclaim from inside dmz_map() This patch avoids that lockdep reports the following: ====================================================== WARNING: possible circular locking dependency detected 4.18.0-rc1 #62 Not tainted ------------------------------------------------------ kswapd0/84 is trying to acquire lock: 00000000c313516d (&xfs_nondir_ilock_class){++++}, at: xfs_free_eofblocks+0xa2/0x1e0 but task is already holding lock: 00000000591c83ae (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x5/0x30 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (fs_reclaim){+.+.}: kmem_cache_alloc+0x2c/0x2b0 radix_tree_node_alloc.constprop.19+0x3d/0xc0 __radix_tree_create+0x161/0x1c0 __radix_tree_insert+0x45/0x210 dmz_map+0x245/0x2d0 [dm_zoned] __map_bio+0x40/0x260 __split_and_process_non_flush+0x116/0x220 __split_and_process_bio+0x81/0x180 __dm_make_request.isra.32+0x5a/0x100 generic_make_request+0x36e/0x690 submit_bio+0x6c/0x140 mpage_readpages+0x19e/0x1f0 read_pages+0x6d/0x1b0 __do_page_cache_readahead+0x21b/0x2d0 force_page_cache_readahead+0xc4/0x100 generic_file_read_iter+0x7c6/0xd20 __vfs_read+0x102/0x180 vfs_read+0x9b/0x140 ksys_read+0x55/0xc0 do_syscall_64+0x5a/0x1f0 entry_SYSCALL_64_after_hwframe+0x49/0xbe -> #1 (&dmz->chunk_lock){+.+.}: dmz_map+0x133/0x2d0 [dm_zoned] __map_bio+0x40/0x260 __split_and_process_non_flush+0x116/0x220 __split_and_process_bio+0x81/0x180 __dm_make_request.isra.32+0x5a/0x100 generic_make_request+0x36e/0x690 submit_bio+0x6c/0x140 _xfs_buf_ioapply+0x31c/0x590 xfs_buf_submit_wait+0x73/0x520 xfs_buf_read_map+0x134/0x2f0 xfs_trans_read_buf_map+0xc3/0x580 xfs_read_agf+0xa5/0x1e0 xfs_alloc_read_agf+0x59/0x2b0 xfs_alloc_pagf_init+0x27/0x60 xfs_bmap_longest_free_extent+0x43/0xb0 xfs_bmap_btalloc_nullfb+0x7f/0xf0 xfs_bmap_btalloc+0x428/0x7c0 xfs_bmapi_write+0x598/0xcc0 xfs_iomap_write_allocate+0x15a/0x330 xfs_map_blocks+0x1cf/0x3f0 xfs_do_writepage+0x15f/0x7b0 write_cache_pages+0x1ca/0x540 xfs_vm_writepages+0x65/0xa0 do_writepages+0x48/0xf0 __writeback_single_inode+0x58/0x730 writeback_sb_inodes+0x249/0x5c0 wb_writeback+0x11e/0x550 wb_workfn+0xa3/0x670 process_one_work+0x228/0x670 worker_thread+0x3c/0x390 kthread+0x11c/0x140 ret_from_fork+0x3a/0x50 -> #0 (&xfs_nondir_ilock_class){++++}: down_read_nested+0x43/0x70 xfs_free_eofblocks+0xa2/0x1e0 xfs_fs_destroy_inode+0xac/0x270 dispose_list+0x51/0x80 prune_icache_sb+0x52/0x70 super_cache_scan+0x127/0x1a0 shrink_slab.part.47+0x1bd/0x590 shrink_node+0x3b5/0x470 balance_pgdat+0x158/0x3b0 kswapd+0x1ba/0x600 kthread+0x11c/0x140 ret_from_fork+0x3a/0x50 other info that might help us debug this: Chain exists of: &xfs_nondir_ilock_class --> &dmz->chunk_lock --> fs_reclaim Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(fs_reclaim); lock(&dmz->chunk_lock); lock(fs_reclaim); lock(&xfs_nondir_ilock_class); *** DEADLOCK *** 3 locks held by kswapd0/84: #0: 00000000591c83ae (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x5/0x30 #1: 000000000f8208f5 (shrinker_rwsem){++++}, at: shrink_slab.part.47+0x3f/0x590 #2: 00000000cacefa54 (&type->s_umount_key#43){.+.+}, at: trylock_super+0x16/0x50 stack backtrace: CPU: 7 PID: 84 Comm: kswapd0 Not tainted 4.18.0-rc1 #62 Hardware name: Supermicro Super Server/X10SRL-F, BIOS 2.0 12/17/2015 Call Trace: dump_stack+0x85/0xcb print_circular_bug.isra.36+0x1ce/0x1db __lock_acquire+0x124e/0x1310 lock_acquire+0x9f/0x1f0 down_read_nested+0x43/0x70 xfs_free_eofblocks+0xa2/0x1e0 xfs_fs_destroy_inode+0xac/0x270 dispose_list+0x51/0x80 prune_icache_sb+0x52/0x70 super_cache_scan+0x127/0x1a0 shrink_slab.part.47+0x1bd/0x590 shrink_node+0x3b5/0x470 balance_pgdat+0x158/0x3b0 kswapd+0x1ba/0x600 kthread+0x11c/0x140 ret_from_fork+0x3a/0x50 Reported-by: Masato Suzuki <masato.suzuki@wdc.com> Fixes: 4218a9554653 ("dm zoned: use GFP_NOIO in I/O path") Cc: <stable@vger.kernel.org> Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-06-22 15:09:11 +00:00
INIT_RADIX_TREE(&dmz->chunk_rxtree, GFP_NOIO);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dmz->chunk_wq = alloc_workqueue("dmz_cwq_%s", WQ_MEM_RECLAIM | WQ_UNBOUND,
0, dev->name);
if (!dmz->chunk_wq) {
ti->error = "Create chunk workqueue failed";
ret = -ENOMEM;
goto err_bio;
}
/* Flush work */
spin_lock_init(&dmz->flush_lock);
bio_list_init(&dmz->flush_list);
INIT_DELAYED_WORK(&dmz->flush_work, dmz_flush_work);
dmz->flush_wq = alloc_ordered_workqueue("dmz_fwq_%s", WQ_MEM_RECLAIM,
dev->name);
if (!dmz->flush_wq) {
ti->error = "Create flush workqueue failed";
ret = -ENOMEM;
goto err_cwq;
}
mod_delayed_work(dmz->flush_wq, &dmz->flush_work, DMZ_FLUSH_PERIOD);
/* Initialize reclaim */
ret = dmz_ctr_reclaim(dev, dmz->metadata, &dmz->reclaim);
if (ret) {
ti->error = "Zone reclaim initialization failed";
goto err_fwq;
}
dmz_dev_info(dev, "Target device: %llu 512-byte logical sectors (%llu blocks)",
(unsigned long long)ti->len,
(unsigned long long)dmz_sect2blk(ti->len));
return 0;
err_fwq:
destroy_workqueue(dmz->flush_wq);
err_cwq:
destroy_workqueue(dmz->chunk_wq);
err_bio:
dm: backfill missing calls to mutex_destroy() Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-01-06 02:17:20 +00:00
mutex_destroy(&dmz->chunk_lock);
dm: convert to bioset_init()/mempool_init() Convert dm to embedded bio sets. Acked-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-20 22:25:53 +00:00
bioset_exit(&dmz->bio_set);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
err_meta:
dmz_dtr_metadata(dmz->metadata);
err_dev:
dmz_put_zoned_device(ti);
err:
kfree(dmz);
return ret;
}
/*
* Cleanup target.
*/
static void dmz_dtr(struct dm_target *ti)
{
struct dmz_target *dmz = ti->private;
flush_workqueue(dmz->chunk_wq);
destroy_workqueue(dmz->chunk_wq);
dmz_dtr_reclaim(dmz->reclaim);
cancel_delayed_work_sync(&dmz->flush_work);
destroy_workqueue(dmz->flush_wq);
(void) dmz_flush_metadata(dmz->metadata);
dmz_dtr_metadata(dmz->metadata);
dm: convert to bioset_init()/mempool_init() Convert dm to embedded bio sets. Acked-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
2018-05-20 22:25:53 +00:00
bioset_exit(&dmz->bio_set);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dmz_put_zoned_device(ti);
dm: backfill missing calls to mutex_destroy() Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-01-06 02:17:20 +00:00
mutex_destroy(&dmz->chunk_lock);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
kfree(dmz);
}
/*
* Setup target request queue limits.
*/
static void dmz_io_hints(struct dm_target *ti, struct queue_limits *limits)
{
struct dmz_target *dmz = ti->private;
unsigned int chunk_sectors = dmz->dev->zone_nr_sectors;
limits->logical_block_size = DMZ_BLOCK_SIZE;
limits->physical_block_size = DMZ_BLOCK_SIZE;
blk_limits_io_min(limits, DMZ_BLOCK_SIZE);
blk_limits_io_opt(limits, DMZ_BLOCK_SIZE);
limits->discard_alignment = DMZ_BLOCK_SIZE;
limits->discard_granularity = DMZ_BLOCK_SIZE;
limits->max_discard_sectors = chunk_sectors;
limits->max_hw_discard_sectors = chunk_sectors;
limits->max_write_zeroes_sectors = chunk_sectors;
/* FS hint to try to align to the device zone size */
limits->chunk_sectors = chunk_sectors;
limits->max_sectors = chunk_sectors;
/* We are exposing a drive-managed zoned block device */
limits->zoned = BLK_ZONED_NONE;
}
/*
* Pass on ioctl to the backend device.
*/
dm: remove fmode_t argument from .prepare_ioctl hook Use the fmode_t that is passed to dm_blk_ioctl() rather than inconsistently (varies across targets) drop it on the floor by overriding it with the fmode_t stored in 'struct dm_dev'. All the persistent reservation functions weren't using the fmode_t they got back from .prepare_ioctl so remove them. Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2018-04-03 20:54:10 +00:00
static int dmz_prepare_ioctl(struct dm_target *ti, struct block_device **bdev)
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
{
struct dmz_target *dmz = ti->private;
*bdev = dmz->dev->bdev;
return 0;
}
/*
* Stop works on suspend.
*/
static void dmz_suspend(struct dm_target *ti)
{
struct dmz_target *dmz = ti->private;
flush_workqueue(dmz->chunk_wq);
dmz_suspend_reclaim(dmz->reclaim);
cancel_delayed_work_sync(&dmz->flush_work);
}
/*
* Restart works on resume or if suspend failed.
*/
static void dmz_resume(struct dm_target *ti)
{
struct dmz_target *dmz = ti->private;
queue_delayed_work(dmz->flush_wq, &dmz->flush_work, DMZ_FLUSH_PERIOD);
dmz_resume_reclaim(dmz->reclaim);
}
static int dmz_iterate_devices(struct dm_target *ti,
iterate_devices_callout_fn fn, void *data)
{
struct dmz_target *dmz = ti->private;
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
struct dmz_dev *dev = dmz->dev;
sector_t capacity = dev->capacity & ~(dev->zone_nr_sectors - 1);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
dm zoned: ignore last smaller runt zone The SCSI layer allows ZBC drives to have a smaller last runt zone. For such a device, specifying the entire capacity for a dm-zoned target table entry fails because the specified capacity is not aligned on a device zone size indicated in the request queue structure of the device. Fix this problem by ignoring the last runt zone in the entry length when seting up the dm-zoned target (ctr method) and when iterating table entries of the target (iterate_devices method). This allows dm-zoned users to still easily setup a target using the entire device capacity (as mandated by dm-zoned) or the aligned capacity excluding the last runt zone. While at it, replace direct references to the device queue chunk_sectors limit with calls to the accessor blk_queue_zone_sectors(). Reported-by: Peter Desnoyers <pjd@ccs.neu.edu> Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-10-28 07:39:34 +00:00
return fn(ti, dmz->ddev, 0, capacity, data);
dm zoned: drive-managed zoned block device target The dm-zoned device mapper target provides transparent write access to zoned block devices (ZBC and ZAC compliant block devices). dm-zoned hides to the device user (a file system or an application doing raw block device accesses) any constraint imposed on write requests by the device, equivalent to a drive-managed zoned block device model. Write requests are processed using a combination of on-disk buffering using the device conventional zones and direct in-place processing for requests aligned to a zone sequential write pointer position. A background reclaim process implemented using dm_kcopyd_copy ensures that conventional zones are always available for executing unaligned write requests. The reclaim process overhead is minimized by managing buffer zones in a least-recently-written order and first targeting the oldest buffer zones. Doing so, blocks under regular write access (such as metadata blocks of a file system) remain stored in conventional zones, resulting in no apparent overhead. dm-zoned implementation focus on simplicity and on minimizing overhead (CPU, memory and storage overhead). For a 14TB host-managed disk with 256 MB zones, dm-zoned memory usage per disk instance is at most about 3 MB and as little as 5 zones will be used internally for storing metadata and performing buffer zone reclaim operations. This is achieved using zone level indirection rather than a full block indirection system for managing block movement between zones. dm-zoned primary target is host-managed zoned block devices but it can also be used with host-aware device models to mitigate potential device-side performance degradation due to excessive random writing. Zoned block devices can be formatted and checked for use with the dm-zoned target using the dmzadm utility available at: https://github.com/hgst/dm-zoned-tools Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com> Reviewed-by: Hannes Reinecke <hare@suse.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [Mike Snitzer partly refactored Damien's original work to cleanup the code] Signed-off-by: Mike Snitzer <snitzer@redhat.com>
2017-06-07 06:55:39 +00:00
}
static struct target_type dmz_type = {
.name = "zoned",
.version = {1, 0, 0},
.features = DM_TARGET_SINGLETON | DM_TARGET_ZONED_HM,
.module = THIS_MODULE,
.ctr = dmz_ctr,
.dtr = dmz_dtr,
.map = dmz_map,
.io_hints = dmz_io_hints,
.prepare_ioctl = dmz_prepare_ioctl,
.postsuspend = dmz_suspend,
.resume = dmz_resume,
.iterate_devices = dmz_iterate_devices,
};
static int __init dmz_init(void)
{
return dm_register_target(&dmz_type);
}
static void __exit dmz_exit(void)
{
dm_unregister_target(&dmz_type);
}
module_init(dmz_init);
module_exit(dmz_exit);
MODULE_DESCRIPTION(DM_NAME " target for zoned block devices");
MODULE_AUTHOR("Damien Le Moal <damien.lemoal@wdc.com>");
MODULE_LICENSE("GPL");
Reference in New Issue Copy Permalink