// SPDX-License-Identifier: GPL-2.0-or-later/* md.c : Multiple Devices driver for Linux Copyright (C) 1998, 1999, 2000 Ingo Molnar completely rewritten, based on the MD driver code from Marc Zyngier Changes: - RAID-1/RAID-5 extensions by Miguel de Icaza, Gadi Oxman, Ingo Molnar - RAID-6 extensions by H. Peter Anvin <hpa@zytor.com> - boot support for linear and striped mode by Harald Hoyer <HarryH@Royal.Net> - kerneld support by Boris Tobotras <boris@xtalk.msk.su> - kmod support by: Cyrus Durgin - RAID0 bugfixes: Mark Anthony Lisher <markal@iname.com> - Devfs support by Richard Gooch <rgooch@atnf.csiro.au> - lots of fixes and improvements to the RAID1/RAID5 and generic RAID code (such as request based resynchronization): Neil Brown <neilb@cse.unsw.edu.au>. - persistent bitmap code Copyright (C) 2003-2004, Paul Clements, SteelEye Technology, Inc. Errors, Warnings, etc. Please use: pr_crit() for error conditions that risk data loss pr_err() for error conditions that are unexpected, like an IO error or internal inconsistency pr_warn() for error conditions that could have been predicated, like adding a device to an array when it has incompatible metadata pr_info() for every interesting, very rare events, like an array starting or stopping, or resync starting or stopping pr_debug() for everything else. */ #include <linux/sched/mm.h>#include <linux/sched/signal.h>#include <linux/kthread.h>#include <linux/blkdev.h>#include <linux/blk-integrity.h>#include <linux/badblocks.h>#include <linux/sysctl.h>#include <linux/seq_file.h>#include <linux/fs.h>#include <linux/poll.h>#include <linux/ctype.h>#include <linux/string.h>#include <linux/hdreg.h>#include <linux/proc_fs.h>#include <linux/random.h>#include <linux/major.h>#include <linux/module.h>#include <linux/reboot.h>#include <linux/file.h>#include <linux/compat.h>#include <linux/delay.h>#include <linux/raid/md_p.h>#include <linux/raid/md_u.h>#include <linux/raid/detect.h>#include <linux/slab.h>#include <linux/percpu-refcount.h>#include <linux/part_stat.h> #include "md.h"#include "md-bitmap.h"#include "md-cluster.h" static const char *action_name[NR_SYNC_ACTIONS] = { [ACTION_RESYNC] = "resync", [ACTION_RECOVER] = "recover", [ACTION_CHECK] = "check", [ACTION_REPAIR] = "repair", [ACTION_RESHAPE] = "reshape", [ACTION_FROZEN] = "frozen", [ACTION_IDLE] = "idle",}; static DEFINE_XARRAY(md_submodule); static const struct kobj_type md_ktype; static DECLARE_WAIT_QUEUE_HEAD(resync_wait); /* * This workqueue is used for sync_work to register new sync_thread, and for * del_work to remove rdev, and for event_work that is only set by dm-raid. * * Noted that sync_work will grab reconfig_mutex, hence never flush this * workqueue whith reconfig_mutex grabbed. */static struct workqueue_struct *md_misc_wq; static int remove_and_add_spares(struct mddev *mddev, struct md_rdev *this);static void mddev_detach(struct mddev *mddev);static void export_rdev(struct md_rdev *rdev);static void md_wakeup_thread_directly(struct md_thread __rcu **thread); /* * Default number of read corrections we'll attempt on an rdev * before ejecting it from the array. We divide the read error * count by 2 for every hour elapsed between read errors. */#define MD_DEFAULT_MAX_CORRECTED_READ_ERRORS 20/* Default safemode delay: 200 msec */#define DEFAULT_SAFEMODE_DELAY ((200 * HZ)/1000 +1)/* * Current RAID-1,4,5,6,10 parallel reconstruction 'guaranteed speed limit' * is sysctl_speed_limit_min, 1000 KB/sec by default, so the extra system load * does not show up that much. Increase it if you want to have more guaranteed * speed. Note that the RAID driver will use the maximum bandwidth * sysctl_speed_limit_max, 200 MB/sec by default, if the IO subsystem is idle. * * Background sync IO speed control: * * - below speed min: * no limit; * - above speed min and below speed max: * a) if mddev is idle, then no limit; * b) if mddev is busy handling normal IO, then limit inflight sync IO * to sync_io_depth; * - above speed max: * sync IO can't be issued; * * Following configurations can be changed via /proc/sys/dev/raid/ for system * or /sys/block/mdX/md/ for one array. */static int sysctl_speed_limit_min = 1000;static int sysctl_speed_limit_max = 200000;static int sysctl_sync_io_depth = 32; static int speed_min(struct mddev *mddev){ return mddev->sync_speed_min ? mddev->sync_speed_min : sysctl_speed_limit_min;} static int speed_max(struct mddev *mddev){ return mddev->sync_speed_max ? mddev->sync_speed_max : sysctl_speed_limit_max;} static int sync_io_depth(struct mddev *mddev){ return mddev->sync_io_depth ? mddev->sync_io_depth : sysctl_sync_io_depth;} static void rdev_uninit_serial(struct md_rdev *rdev){ if (!test_and_clear_bit(CollisionCheck, &rdev->flags)) return; kvfree(rdev->serial); rdev->serial = NULL;} static void rdevs_uninit_serial(struct mddev *mddev){ struct md_rdev *rdev; rdev_for_each(rdev, mddev) rdev_uninit_serial(rdev);} static int rdev_init_serial(struct md_rdev *rdev){ /* serial_nums equals with BARRIER_BUCKETS_NR */ int i, serial_nums = 1 << ((PAGE_SHIFT - ilog2(sizeof(atomic_t)))); struct serial_in_rdev *serial = NULL; if (test_bit(CollisionCheck, &rdev->flags)) return 0; serial = kvmalloc(sizeof(struct serial_in_rdev) * serial_nums, GFP_KERNEL); if (!serial) return -ENOMEM; for (i = 0; i < serial_nums; i++) { struct serial_in_rdev *serial_tmp = &serial[i]; spin_lock_init(&serial_tmp->serial_lock); serial_tmp->serial_rb = RB_ROOT_CACHED; } rdev->serial = serial; set_bit(CollisionCheck, &rdev->flags); return 0;} static int rdevs_init_serial(struct mddev *mddev){ struct md_rdev *rdev; int ret = 0; rdev_for_each(rdev, mddev) { ret = rdev_init_serial(rdev); if (ret) break; } /* Free all resources if pool is not existed */ if (ret && !mddev->serial_info_pool) rdevs_uninit_serial(mddev); return ret;} /* * rdev needs to enable serial stuffs if it meets the conditions: * 1. it is multi-queue device flaged with writemostly. * 2. the write-behind mode is enabled. */static int rdev_need_serial(struct md_rdev *rdev){ return (rdev && rdev->mddev->bitmap_info.max_write_behind > 0 && rdev->bdev->bd_disk->queue->nr_hw_queues != 1 && test_bit(WriteMostly, &rdev->flags));} /* * Init resource for rdev(s), then create serial_info_pool if: * 1. rdev is the first device which return true from rdev_enable_serial. * 2. rdev is NULL, means we want to enable serialization for all rdevs. */void mddev_create_serial_pool(struct mddev *mddev, struct md_rdev *rdev){ int ret = 0; unsigned int noio_flags; if (!test_bit(MD_SERIALIZE_POLICY, &mddev->flags) && rdev && !rdev_need_serial(rdev) && !test_bit(CollisionCheck, &rdev->flags)) return; noio_flags = memalloc_noio_save(); if (!rdev) ret = rdevs_init_serial(mddev); else ret = rdev_init_serial(rdev); if (ret) goto out; if (mddev->serial_info_pool == NULL) { mddev->serial_info_pool = mempool_create_kmalloc_pool(NR_SERIAL_INFOS, sizeof(struct serial_info)); if (!mddev->serial_info_pool) { rdevs_uninit_serial(mddev); pr_err("can't alloc memory pool for serialization\n"); } }out: memalloc_noio_restore(noio_flags);} /* * Free resource from rdev(s), and destroy serial_info_pool under conditions: * 1. rdev is the last device flaged with CollisionCheck. * 2. when bitmap is destroyed while policy is not enabled. * 3. for disable policy, the pool is destroyed only when no rdev needs it. */void mddev_destroy_serial_pool(struct mddev *mddev, struct md_rdev *rdev){ if (rdev && !test_bit(CollisionCheck, &rdev->flags)) return; if (mddev->serial_info_pool) { struct md_rdev *temp; int num = 0; /* used to track if other rdevs need the pool */ rdev_for_each(temp, mddev) { if (!rdev) { if (!test_bit(MD_SERIALIZE_POLICY, &mddev->flags) || !rdev_need_serial(temp)) rdev_uninit_serial(temp); else num++; } else if (temp != rdev && test_bit(CollisionCheck, &temp->flags)) num++; } if (rdev) rdev_uninit_serial(rdev); if (num) pr_info("The mempool could be used by other devices\n"); else { mempool_destroy(mddev->serial_info_pool); mddev->serial_info_pool = NULL; } }} static struct ctl_table_header *raid_table_header; static const struct ctl_table raid_table[] = { { .procname = "speed_limit_min", .data = &sysctl_speed_limit_min, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, }, { .procname = "speed_limit_max", .data = &sysctl_speed_limit_max, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, }, { .procname = "sync_io_depth", .data = &sysctl_sync_io_depth, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, },}; static int start_readonly; /* * The original mechanism for creating an md device is to create * a device node in /dev and to open it. This causes races with device-close. * The preferred method is to write to the "new_array" module parameter. * This can avoid races. * Setting create_on_open to false disables the original mechanism * so all the races disappear. */static bool create_on_open = true;static bool legacy_async_del_gendisk = true;static bool check_new_feature = true; /* * We have a system wide 'event count' that is incremented * on any 'interesting' event, and readers of /proc/mdstat * can use 'poll' or 'select' to find out when the event * count increases. * * Events are: * start array, stop array, error, add device, remove device, * start build, activate spare */static DECLARE_WAIT_QUEUE_HEAD(md_event_waiters);static atomic_t md_event_count;void md_new_event(void){ atomic_inc(&md_event_count); wake_up(&md_event_waiters);}EXPORT_SYMBOL_GPL(md_new_event); /* * Enables to iterate over all existing md arrays * all_mddevs_lock protects this list. */static LIST_HEAD(all_mddevs);static DEFINE_SPINLOCK(all_mddevs_lock); static bool is_md_suspended(struct mddev *mddev){ return percpu_ref_is_dying(&mddev->active_io);}/* Rather than calling directly into the personality make_request function, * IO requests come here first so that we can check if the device is * being suspended pending a reconfiguration. * We hold a refcount over the call to ->make_request. By the time that * call has finished, the bio has been linked into some internal structure * and so is visible to ->quiesce(), so we don't need the refcount any more. */static bool is_suspended(struct mddev *mddev, struct bio *bio){ if (is_md_suspended(mddev)) return true; if (bio_data_dir(bio) != WRITE) return false; if (READ_ONCE(mddev->suspend_lo) >= READ_ONCE(mddev->suspend_hi)) return false; if (bio->bi_iter.bi_sector >= READ_ONCE(mddev->suspend_hi)) return false; if (bio_end_sector(bio) < READ_ONCE(mddev->suspend_lo)) return false; return true;} bool md_handle_request(struct mddev *mddev, struct bio *bio){check_suspended: if (unlikely(md_cloned_bio(mddev, bio))) { /* * This bio is an MD cloned bio and already holds an * active_io reference, so percpu_ref_get() is safe here. */ percpu_ref_get(&mddev->active_io); } else { if (is_suspended(mddev, bio)) { /* Bail out if REQ_NOWAIT is set for the bio */ if (bio->bi_opf & REQ_NOWAIT) { bio_wouldblock_error(bio); return true; } wait_event(mddev->sb_wait, !is_suspended(mddev, bio)); } if (!percpu_ref_tryget_live(&mddev->active_io)) goto check_suspended; } if (!mddev->pers->make_request(mddev, bio)) { percpu_ref_put(&mddev->active_io); if (mddev_is_dm(mddev) && mddev->pers->prepare_suspend) return false; goto check_suspended; } percpu_ref_put(&mddev->active_io); return true;}EXPORT_SYMBOL(md_handle_request); static void md_submit_bio(struct bio *bio){ const int rw = bio_data_dir(bio); struct mddev *mddev = bio->bi_bdev->bd_disk->private_data; if (mddev == NULL || mddev->pers == NULL) { bio_io_error(bio); return; } if (unlikely(test_bit(MD_BROKEN, &mddev->flags)) && (rw == WRITE)) { bio_io_error(bio); return; } bio = bio_split_to_limits(bio); if (!bio) return; if (mddev->ro == MD_RDONLY && unlikely(rw == WRITE)) { if (bio_sectors(bio) != 0) bio->bi_status = BLK_STS_IOERR; bio_endio(bio); return; } /* bio could be mergeable after passing to underlayer */ bio->bi_opf &= ~REQ_NOMERGE; md_handle_request(mddev, bio);} /* * Make sure no new requests are submitted to the device, and any requests that * have been submitted are completely handled. */int mddev_suspend(struct mddev *mddev, bool interruptible){ int err = 0; /* * hold reconfig_mutex to wait for normal io will deadlock, because * other context can't update super_block, and normal io can rely on * updating super_block. */ lockdep_assert_not_held(&mddev->reconfig_mutex); if (interruptible) err = mutex_lock_interruptible(&mddev->suspend_mutex); else mutex_lock(&mddev->suspend_mutex); if (err) return err; if (mddev->suspended) { WRITE_ONCE(mddev->suspended, mddev->suspended + 1); mutex_unlock(&mddev->suspend_mutex); return 0; } percpu_ref_kill(&mddev->active_io); /* * RAID456 IO can sleep in wait_for_reshape while still holding an * active_io reference. If reshape is already interrupted or frozen, * wake those waiters so they can abort and drop the reference instead * of deadlocking suspend. */ if (mddev->pers && mddev->pers->prepare_suspend && reshape_interrupted(mddev)) mddev->pers->prepare_suspend(mddev); Lines 1–500 of 11121. Load more
// SPDX-License-Identifier: GPL-2.0-or-later /* md.c : Multiple Devices driver for Linux Copyright (C) 1998, 1999, 2000 Ingo Molnar completely rewritten, based on the MD driver code from Marc Zyngier Changes: - RAID-1/RAID-5 extensions by Miguel de Icaza, Gadi Oxman, Ingo Molnar - RAID-6 extensions by H. Peter Anvin <hpa@zytor.com> - boot support for linear and striped mode by Harald Hoyer <HarryH@Royal.Net> - kerneld support by Boris Tobotras <boris@xtalk.msk.su> - kmod support by: Cyrus Durgin - RAID0 bugfixes: Mark Anthony Lisher <markal@iname.com> - Devfs support by Richard Gooch <rgooch@atnf.csiro.au> - lots of fixes and improvements to the RAID1/RAID5 and generic RAID code (such as request based resynchronization): Neil Brown <neilb@cse.unsw.edu.au>. - persistent bitmap code Copyright (C) 2003-2004, Paul Clements, SteelEye Technology, Inc. Errors, Warnings, etc. Please use: pr_crit() for error conditions that risk data loss pr_err() for error conditions that are unexpected, like an IO error or internal inconsistency pr_warn() for error conditions that could have been predicated, like adding a device to an array when it has incompatible metadata pr_info() for every interesting, very rare events, like an array starting or stopping, or resync starting or stopping pr_debug() for everything else. */ #include <linux/sched/mm.h> #include <linux/sched/signal.h> #include <linux/kthread.h> #include <linux/blkdev.h> #include <linux/blk-integrity.h> #include <linux/badblocks.h> #include <linux/sysctl.h> #include <linux/seq_file.h> #include <linux/fs.h> #include <linux/poll.h> #include <linux/ctype.h> #include <linux/string.h> #include <linux/hdreg.h> #include <linux/proc_fs.h> #include <linux/random.h> #include <linux/major.h> #include <linux/module.h> #include <linux/reboot.h> #include <linux/file.h> #include <linux/compat.h> #include <linux/delay.h> #include <linux/raid/md_p.h> #include <linux/raid/md_u.h> #include <linux/raid/detect.h> #include <linux/slab.h> #include <linux/percpu-refcount.h> #include <linux/part_stat.h> #include "md.h" #include "md-bitmap.h" #include "md-cluster.h" static const char *action_name[NR_SYNC_ACTIONS] = { [ACTION_RESYNC] = "resync", [ACTION_RECOVER] = "recover", [ACTION_CHECK] = "check", [ACTION_REPAIR] = "repair", [ACTION_RESHAPE] = "reshape", [ACTION_FROZEN] = "frozen", [ACTION_IDLE] = "idle", }; static DEFINE_XARRAY(md_submodule); static const struct kobj_type md_ktype; static DECLARE_WAIT_QUEUE_HEAD(resync_wait); /* * This workqueue is used for sync_work to register new sync_thread, and for * del_work to remove rdev, and for event_work that is only set by dm-raid. * * Noted that sync_work will grab reconfig_mutex, hence never flush this * workqueue whith reconfig_mutex grabbed. */ static struct workqueue_struct *md_misc_wq; static int remove_and_add_spares(struct mddev *mddev, struct md_rdev *this); static void mddev_detach(struct mddev *mddev); static void export_rdev(struct md_rdev *rdev); static void md_wakeup_thread_directly(struct md_thread __rcu **thread); /* * Default number of read corrections we'll attempt on an rdev * before ejecting it from the array. We divide the read error * count by 2 for every hour elapsed between read errors. */ #define MD_DEFAULT_MAX_CORRECTED_READ_ERRORS 20 /* Default safemode delay: 200 msec */ #define DEFAULT_SAFEMODE_DELAY ((200 * HZ)/1000 +1) /* * Current RAID-1,4,5,6,10 parallel reconstruction 'guaranteed speed limit' * is sysctl_speed_limit_min, 1000 KB/sec by default, so the extra system load * does not show up that much. Increase it if you want to have more guaranteed * speed. Note that the RAID driver will use the maximum bandwidth * sysctl_speed_limit_max, 200 MB/sec by default, if the IO subsystem is idle. * * Background sync IO speed control: * * - below speed min: * no limit; * - above speed min and below speed max: * a) if mddev is idle, then no limit; * b) if mddev is busy handling normal IO, then limit inflight sync IO * to sync_io_depth; * - above speed max: * sync IO can't be issued; * * Following configurations can be changed via /proc/sys/dev/raid/ for system * or /sys/block/mdX/md/ for one array. */ static int sysctl_speed_limit_min = 1000; static int sysctl_speed_limit_max = 200000; static int sysctl_sync_io_depth = 32; static int speed_min(struct mddev *mddev) { return mddev->sync_speed_min ? mddev->sync_speed_min : sysctl_speed_limit_min; } static int speed_max(struct mddev *mddev) { return mddev->sync_speed_max ? mddev->sync_speed_max : sysctl_speed_limit_max; } static int sync_io_depth(struct mddev *mddev) { return mddev->sync_io_depth ? mddev->sync_io_depth : sysctl_sync_io_depth; } static void rdev_uninit_serial(struct md_rdev *rdev) { if (!test_and_clear_bit(CollisionCheck, &rdev->flags)) return; kvfree(rdev->serial); rdev->serial = NULL; } static void rdevs_uninit_serial(struct mddev *mddev) { struct md_rdev *rdev; rdev_for_each(rdev, mddev) rdev_uninit_serial(rdev); } static int rdev_init_serial(struct md_rdev *rdev) { /* serial_nums equals with BARRIER_BUCKETS_NR */ int i, serial_nums = 1 << ((PAGE_SHIFT - ilog2(sizeof(atomic_t)))); struct serial_in_rdev *serial = NULL; if (test_bit(CollisionCheck, &rdev->flags)) return 0; serial = kvmalloc(sizeof(struct serial_in_rdev) * serial_nums, GFP_KERNEL); if (!serial) return -ENOMEM; for (i = 0; i < serial_nums; i++) { struct serial_in_rdev *serial_tmp = &serial[i]; spin_lock_init(&serial_tmp->serial_lock); serial_tmp->serial_rb = RB_ROOT_CACHED; } rdev->serial = serial; set_bit(CollisionCheck, &rdev->flags); return 0; } static int rdevs_init_serial(struct mddev *mddev) { struct md_rdev *rdev; int ret = 0; rdev_for_each(rdev, mddev) { ret = rdev_init_serial(rdev); if (ret) break; } /* Free all resources if pool is not existed */ if (ret && !mddev->serial_info_pool) rdevs_uninit_serial(mddev); return ret; } /* * rdev needs to enable serial stuffs if it meets the conditions: * 1. it is multi-queue device flaged with writemostly. * 2. the write-behind mode is enabled. */ static int rdev_need_serial(struct md_rdev *rdev) { return (rdev && rdev->mddev->bitmap_info.max_write_behind > 0 && rdev->bdev->bd_disk->queue->nr_hw_queues != 1 && test_bit(WriteMostly, &rdev->flags)); } /* * Init resource for rdev(s), then create serial_info_pool if: * 1. rdev is the first device which return true from rdev_enable_serial. * 2. rdev is NULL, means we want to enable serialization for all rdevs. */ void mddev_create_serial_pool(struct mddev *mddev, struct md_rdev *rdev) { int ret = 0; unsigned int noio_flags; if (!test_bit(MD_SERIALIZE_POLICY, &mddev->flags) && rdev && !rdev_need_serial(rdev) && !test_bit(CollisionCheck, &rdev->flags)) return; noio_flags = memalloc_noio_save(); if (!rdev) ret = rdevs_init_serial(mddev); else ret = rdev_init_serial(rdev); if (ret) goto out; if (mddev->serial_info_pool == NULL) { mddev->serial_info_pool = mempool_create_kmalloc_pool(NR_SERIAL_INFOS, sizeof(struct serial_info)); if (!mddev->serial_info_pool) { rdevs_uninit_serial(mddev); pr_err("can't alloc memory pool for serialization\n"); } } out: memalloc_noio_restore(noio_flags); } /* * Free resource from rdev(s), and destroy serial_info_pool under conditions: * 1. rdev is the last device flaged with CollisionCheck. * 2. when bitmap is destroyed while policy is not enabled. * 3. for disable policy, the pool is destroyed only when no rdev needs it. */ void mddev_destroy_serial_pool(struct mddev *mddev, struct md_rdev *rdev) { if (rdev && !test_bit(CollisionCheck, &rdev->flags)) return; if (mddev->serial_info_pool) { struct md_rdev *temp; int num = 0; /* used to track if other rdevs need the pool */ rdev_for_each(temp, mddev) { if (!rdev) { if (!test_bit(MD_SERIALIZE_POLICY, &mddev->flags) || !rdev_need_serial(temp)) rdev_uninit_serial(temp); else num++; } else if (temp != rdev && test_bit(CollisionCheck, &temp->flags)) num++; } if (rdev) rdev_uninit_serial(rdev); if (num) pr_info("The mempool could be used by other devices\n"); else { mempool_destroy(mddev->serial_info_pool); mddev->serial_info_pool = NULL; } } } static struct ctl_table_header *raid_table_header; static const struct ctl_table raid_table[] = { { .procname = "speed_limit_min", .data = &sysctl_speed_limit_min, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, }, { .procname = "speed_limit_max", .data = &sysctl_speed_limit_max, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, }, { .procname = "sync_io_depth", .data = &sysctl_sync_io_depth, .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dointvec, }, }; static int start_readonly; /* * The original mechanism for creating an md device is to create * a device node in /dev and to open it. This causes races with device-close. * The preferred method is to write to the "new_array" module parameter. * This can avoid races. * Setting create_on_open to false disables the original mechanism * so all the races disappear. */ static bool create_on_open = true; static bool legacy_async_del_gendisk = true; static bool check_new_feature = true; /* * We have a system wide 'event count' that is incremented * on any 'interesting' event, and readers of /proc/mdstat * can use 'poll' or 'select' to find out when the event * count increases. * * Events are: * start array, stop array, error, add device, remove device, * start build, activate spare */ static DECLARE_WAIT_QUEUE_HEAD(md_event_waiters); static atomic_t md_event_count; void md_new_event(void) { atomic_inc(&md_event_count); wake_up(&md_event_waiters); } EXPORT_SYMBOL_GPL(md_new_event); /* * Enables to iterate over all existing md arrays * all_mddevs_lock protects this list. */ static LIST_HEAD(all_mddevs); static DEFINE_SPINLOCK(all_mddevs_lock); static bool is_md_suspended(struct mddev *mddev) { return percpu_ref_is_dying(&mddev->active_io); } /* Rather than calling directly into the personality make_request function, * IO requests come here first so that we can check if the device is * being suspended pending a reconfiguration. * We hold a refcount over the call to ->make_request. By the time that * call has finished, the bio has been linked into some internal structure * and so is visible to ->quiesce(), so we don't need the refcount any more. */ static bool is_suspended(struct mddev *mddev, struct bio *bio) { if (is_md_suspended(mddev)) return true; if (bio_data_dir(bio) != WRITE) return false; if (READ_ONCE(mddev->suspend_lo) >= READ_ONCE(mddev->suspend_hi)) return false; if (bio->bi_iter.bi_sector >= READ_ONCE(mddev->suspend_hi)) return false; if (bio_end_sector(bio) < READ_ONCE(mddev->suspend_lo)) return false; return true; } bool md_handle_request(struct mddev *mddev, struct bio *bio) { check_suspended: if (unlikely(md_cloned_bio(mddev, bio))) { /* * This bio is an MD cloned bio and already holds an * active_io reference, so percpu_ref_get() is safe here. */ percpu_ref_get(&mddev->active_io); } else { if (is_suspended(mddev, bio)) { /* Bail out if REQ_NOWAIT is set for the bio */ if (bio->bi_opf & REQ_NOWAIT) { bio_wouldblock_error(bio); return true; } wait_event(mddev->sb_wait, !is_suspended(mddev, bio)); } if (!percpu_ref_tryget_live(&mddev->active_io)) goto check_suspended; } if (!mddev->pers->make_request(mddev, bio)) { percpu_ref_put(&mddev->active_io); if (mddev_is_dm(mddev) && mddev->pers->prepare_suspend) return false; goto check_suspended; } percpu_ref_put(&mddev->active_io); return true; } EXPORT_SYMBOL(md_handle_request); static void md_submit_bio(struct bio *bio) { const int rw = bio_data_dir(bio); struct mddev *mddev = bio->bi_bdev->bd_disk->private_data; if (mddev == NULL || mddev->pers == NULL) { bio_io_error(bio); return; } if (unlikely(test_bit(MD_BROKEN, &mddev->flags)) && (rw == WRITE)) { bio_io_error(bio); return; } bio = bio_split_to_limits(bio); if (!bio) return; if (mddev->ro == MD_RDONLY && unlikely(rw == WRITE)) { if (bio_sectors(bio) != 0) bio->bi_status = BLK_STS_IOERR; bio_endio(bio); return; } /* bio could be mergeable after passing to underlayer */ bio->bi_opf &= ~REQ_NOMERGE; md_handle_request(mddev, bio); } /* * Make sure no new requests are submitted to the device, and any requests that * have been submitted are completely handled. */ int mddev_suspend(struct mddev *mddev, bool interruptible) { int err = 0; /* * hold reconfig_mutex to wait for normal io will deadlock, because * other context can't update super_block, and normal io can rely on * updating super_block. */ lockdep_assert_not_held(&mddev->reconfig_mutex); if (interruptible) err = mutex_lock_interruptible(&mddev->suspend_mutex); else mutex_lock(&mddev->suspend_mutex); if (err) return err; if (mddev->suspended) { WRITE_ONCE(mddev->suspended, mddev->suspended + 1); mutex_unlock(&mddev->suspend_mutex); return 0; } percpu_ref_kill(&mddev->active_io); /* * RAID456 IO can sleep in wait_for_reshape while still holding an * active_io reference. If reshape is already interrupted or frozen, * wake those waiters so they can abort and drop the reference instead * of deadlocking suspend. */ if (mddev->pers && mddev->pers->prepare_suspend && reshape_interrupted(mddev)) mddev->pers->prepare_suspend(mddev);