linux/drivers/md/raid1.h

206 lines
5.7 KiB
C
Raw Normal View History

#ifndef _RAID1_H
#define _RAID1_H
RAID1: a new I/O barrier implementation to remove resync window 'Commit 79ef3a8aa1cb ("raid1: Rewrite the implementation of iobarrier.")' introduces a sliding resync window for raid1 I/O barrier, this idea limits I/O barriers to happen only inside a slidingresync window, for regular I/Os out of this resync window they don't need to wait for barrier any more. On large raid1 device, it helps a lot to improve parallel writing I/O throughput when there are background resync I/Os performing at same time. The idea of sliding resync widow is awesome, but code complexity is a challenge. Sliding resync window requires several variables to work collectively, this is complexed and very hard to make it work correctly. Just grep "Fixes: 79ef3a8aa1" in kernel git log, there are 8 more patches to fix the original resync window patch. This is not the end, any further related modification may easily introduce more regreassion. Therefore I decide to implement a much simpler raid1 I/O barrier, by removing resync window code, I believe life will be much easier. The brief idea of the simpler barrier is, - Do not maintain a global unique resync window - Use multiple hash buckets to reduce I/O barrier conflicts, regular I/O only has to wait for a resync I/O when both them have same barrier bucket index, vice versa. - I/O barrier can be reduced to an acceptable number if there are enough barrier buckets Here I explain how the barrier buckets are designed, - BARRIER_UNIT_SECTOR_SIZE The whole LBA address space of a raid1 device is divided into multiple barrier units, by the size of BARRIER_UNIT_SECTOR_SIZE. Bio requests won't go across border of barrier unit size, that means maximum bio size is BARRIER_UNIT_SECTOR_SIZE<<9 (64MB) in bytes. For random I/O 64MB is large enough for both read and write requests, for sequential I/O considering underlying block layer may merge them into larger requests, 64MB is still good enough. Neil also points out that for resync operation, "we want the resync to move from region to region fairly quickly so that the slowness caused by having to synchronize with the resync is averaged out over a fairly small time frame". For full speed resync, 64MB should take less then 1 second. When resync is competing with other I/O, it could take up a few minutes. Therefore 64MB size is fairly good range for resync. - BARRIER_BUCKETS_NR There are BARRIER_BUCKETS_NR buckets in total, which is defined by, #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - 2) #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) this patch makes the bellowed members of struct r1conf from integer to array of integers, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int *nr_pending; + int *nr_waiting; + int *nr_queued; + int *barrier; number of the array elements is defined as BARRIER_BUCKETS_NR. For 4KB kernel space page size, (PAGE_SHIFT - 2) indecates there are 1024 I/O barrier buckets, and each array of integers occupies single memory page. 1024 means for a request which is smaller than the I/O barrier unit size has ~0.1% chance to wait for resync to pause, which is quite a small enough fraction. Also requesting single memory page is more friendly to kernel page allocator than larger memory size. - I/O barrier bucket is indexed by bio start sector If multiple I/O requests hit different I/O barrier units, they only need to compete I/O barrier with other I/Os which hit the same I/O barrier bucket index with each other. The index of a barrier bucket which a bio should look for is calculated by sector_to_idx() which is defined in raid1.h as an inline function, static inline int sector_to_idx(sector_t sector) { return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, BARRIER_BUCKETS_NR_BITS); } Here sector_nr is the start sector number of a bio. - Single bio won't go across boundary of a I/O barrier unit If a request goes across boundary of barrier unit, it will be split. A bio may be split in raid1_make_request() or raid1_sync_request(), if sectors returned by align_to_barrier_unit_end() is smaller than original bio size. Comparing to single sliding resync window, - Currently resync I/O grows linearly, therefore regular and resync I/O will conflict within a single barrier units. So the I/O behavior is similar to single sliding resync window. - But a barrier unit bucket is shared by all barrier units with identical barrier uinit index, the probability of conflict might be higher than single sliding resync window, in condition that writing I/Os always hit barrier units which have identical barrier bucket indexs with the resync I/Os. This is a very rare condition in real I/O work loads, I cannot imagine how it could happen in practice. - Therefore we can achieve a good enough low conflict rate with much simpler barrier algorithm and implementation. There are two changes should be noticed, - In raid1d(), I change the code to decrease conf->nr_pending[idx] into single loop, it looks like this, spin_lock_irqsave(&conf->device_lock, flags); conf->nr_queued[idx]--; spin_unlock_irqrestore(&conf->device_lock, flags); This change generates more spin lock operations, but in next patch of this patch set, it will be replaced by a single line code, atomic_dec(&conf->nr_queueud[idx]); So we don't need to worry about spin lock cost here. - Mainline raid1 code split original raid1_make_request() into raid1_read_request() and raid1_write_request(). If the original bio goes across an I/O barrier unit size, this bio will be split before calling raid1_read_request() or raid1_write_request(), this change the code logic more simple and clear. - In this patch wait_barrier() is moved from raid1_make_request() to raid1_write_request(). In raid_read_request(), original wait_barrier() is replaced by raid1_read_request(). The differnece is wait_read_barrier() only waits if array is frozen, using different barrier function in different code path makes the code more clean and easy to read. Changelog V4: - Add alloc_r1bio() to remove redundant r1bio memory allocation code. - Fix many typos in patch comments. - Use (PAGE_SHIFT - ilog2(sizeof(int))) to define BARRIER_BUCKETS_NR_BITS. V3: - Rebase the patch against latest upstream kernel code. - Many fixes by review comments from Neil, - Back to use pointers to replace arraries in struct r1conf - Remove total_barriers from struct r1conf - Add more patch comments to explain how/why the values of BARRIER_UNIT_SECTOR_SIZE and BARRIER_BUCKETS_NR are decided. - Use get_unqueued_pending() to replace get_all_pendings() and get_all_queued() - Increase bucket number from 512 to 1024 - Change code comments format by review from Shaohua. V2: - Use bio_split() to split the orignal bio if it goes across barrier unit bounday, to make the code more simple, by suggestion from Shaohua and Neil. - Use hash_long() to replace original linear hash, to avoid a possible confilict between resync I/O and sequential write I/O, by suggestion from Shaohua. - Add conf->total_barriers to record barrier depth, which is used to control number of parallel sync I/O barriers, by suggestion from Shaohua. - In V1 patch the bellowed barrier buckets related members in r1conf are allocated in memory page. To make the code more simple, V2 patch moves the memory space into struct r1conf, like this, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int nr_pending[BARRIER_BUCKETS_NR]; + int nr_waiting[BARRIER_BUCKETS_NR]; + int nr_queued[BARRIER_BUCKETS_NR]; + int barrier[BARRIER_BUCKETS_NR]; This change is by the suggestion from Shaohua. - Remove some inrelavent code comments, by suggestion from Guoqing. - Add a missing wait_barrier() before jumping to retry_write, in raid1_make_write_request(). V1: - Original RFC patch for comments Signed-off-by: Coly Li <colyli@suse.de> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:56 +00:00
/*
* each barrier unit size is 64MB fow now
* note: it must be larger than RESYNC_DEPTH
*/
#define BARRIER_UNIT_SECTOR_BITS 17
#define BARRIER_UNIT_SECTOR_SIZE (1<<17)
/*
* In struct r1conf, the following members are related to I/O barrier
* buckets,
RAID1: avoid unnecessary spin locks in I/O barrier code When I run a parallel reading performan testing on a md raid1 device with two NVMe SSDs, I observe very bad throughput in supprise: by fio with 64KB block size, 40 seq read I/O jobs, 128 iodepth, overall throughput is only 2.7GB/s, this is around 50% of the idea performance number. The perf reports locking contention happens at allow_barrier() and wait_barrier() code, - 41.41% fio [kernel.kallsyms] [k] _raw_spin_lock_irqsave - _raw_spin_lock_irqsave + 89.92% allow_barrier + 9.34% __wake_up - 37.30% fio [kernel.kallsyms] [k] _raw_spin_lock_irq - _raw_spin_lock_irq - 100.00% wait_barrier The reason is, in these I/O barrier related functions, - raise_barrier() - lower_barrier() - wait_barrier() - allow_barrier() They always hold conf->resync_lock firstly, even there are only regular reading I/Os and no resync I/O at all. This is a huge performance penalty. The solution is a lockless-like algorithm in I/O barrier code, and only holding conf->resync_lock when it has to. The original idea is from Hannes Reinecke, and Neil Brown provides comments to improve it. I continue to work on it, and make the patch into current form. In the new simpler raid1 I/O barrier implementation, there are two wait barrier functions, - wait_barrier() Which calls _wait_barrier(), is used for regular write I/O. If there is resync I/O happening on the same I/O barrier bucket, or the whole array is frozen, task will wait until no barrier on same barrier bucket, or the whold array is unfreezed. - wait_read_barrier() Since regular read I/O won't interfere with resync I/O (read_balance() will make sure only uptodate data will be read out), it is unnecessary to wait for barrier in regular read I/Os, waiting in only necessary when the whole array is frozen. The operations on conf->nr_pending[idx], conf->nr_waiting[idx], conf-> barrier[idx] are very carefully designed in raise_barrier(), lower_barrier(), _wait_barrier() and wait_read_barrier(), in order to avoid unnecessary spin locks in these functions. Once conf-> nr_pengding[idx] is increased, a resync I/O with same barrier bucket index has to wait in raise_barrier(). Then in _wait_barrier() if no barrier raised in same barrier bucket index and array is not frozen, the regular I/O doesn't need to hold conf->resync_lock, it can just increase conf->nr_pending[idx], and return to its caller. wait_read_barrier() is very similar to _wait_barrier(), the only difference is it only waits when array is frozen. For heavy parallel reading I/Os, the lockless I/O barrier code almostly gets rid of all spin lock cost. This patch significantly improves raid1 reading peroformance. From my testing, a raid1 device built by two NVMe SSD, runs fio with 64KB blocksize, 40 seq read I/O jobs, 128 iodepth, overall throughput increases from 2.7GB/s to 4.6GB/s (+70%). Changelog V4: - Change conf->nr_queued[] to atomic_t. - Define BARRIER_BUCKETS_NR_BITS by (PAGE_SHIFT - ilog2(sizeof(atomic_t))) V3: - Add smp_mb__after_atomic() as Shaohua and Neil suggested. - Change conf->nr_queued[] from atomic_t to int. - Change conf->array_frozen from atomic_t back to int, and use READ_ONCE(conf->array_frozen) to check value of conf->array_frozen in _wait_barrier() and wait_read_barrier(). - In _wait_barrier() and wait_read_barrier(), add a call to wake_up(&conf->wait_barrier) after atomic_dec(&conf->nr_pending[idx]), to fix a deadlock between _wait_barrier()/wait_read_barrier and freeze_array(). V2: - Remove a spin_lock/unlock pair in raid1d(). - Add more code comments to explain why there is no racy when checking two atomic_t variables at same time. V1: - Original RFC patch for comments. Signed-off-by: Coly Li <colyli@suse.de> Cc: Shaohua Li <shli@fb.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:57 +00:00
* atomic_t *nr_pending;
* atomic_t *nr_waiting;
* atomic_t *nr_queued;
* atomic_t *barrier;
* Each of them points to array of atomic_t variables, each array is
* designed to have BARRIER_BUCKETS_NR elements and occupy a single
* memory page. The data width of atomic_t variables is 4 bytes, equal
* to 1<<(ilog2(sizeof(atomic_t))), BARRIER_BUCKETS_NR_BITS is defined
* as (PAGE_SHIFT - ilog2(sizeof(int))) to make sure an array of
* atomic_t variables with BARRIER_BUCKETS_NR elements just exactly
* occupies a single memory page.
RAID1: a new I/O barrier implementation to remove resync window 'Commit 79ef3a8aa1cb ("raid1: Rewrite the implementation of iobarrier.")' introduces a sliding resync window for raid1 I/O barrier, this idea limits I/O barriers to happen only inside a slidingresync window, for regular I/Os out of this resync window they don't need to wait for barrier any more. On large raid1 device, it helps a lot to improve parallel writing I/O throughput when there are background resync I/Os performing at same time. The idea of sliding resync widow is awesome, but code complexity is a challenge. Sliding resync window requires several variables to work collectively, this is complexed and very hard to make it work correctly. Just grep "Fixes: 79ef3a8aa1" in kernel git log, there are 8 more patches to fix the original resync window patch. This is not the end, any further related modification may easily introduce more regreassion. Therefore I decide to implement a much simpler raid1 I/O barrier, by removing resync window code, I believe life will be much easier. The brief idea of the simpler barrier is, - Do not maintain a global unique resync window - Use multiple hash buckets to reduce I/O barrier conflicts, regular I/O only has to wait for a resync I/O when both them have same barrier bucket index, vice versa. - I/O barrier can be reduced to an acceptable number if there are enough barrier buckets Here I explain how the barrier buckets are designed, - BARRIER_UNIT_SECTOR_SIZE The whole LBA address space of a raid1 device is divided into multiple barrier units, by the size of BARRIER_UNIT_SECTOR_SIZE. Bio requests won't go across border of barrier unit size, that means maximum bio size is BARRIER_UNIT_SECTOR_SIZE<<9 (64MB) in bytes. For random I/O 64MB is large enough for both read and write requests, for sequential I/O considering underlying block layer may merge them into larger requests, 64MB is still good enough. Neil also points out that for resync operation, "we want the resync to move from region to region fairly quickly so that the slowness caused by having to synchronize with the resync is averaged out over a fairly small time frame". For full speed resync, 64MB should take less then 1 second. When resync is competing with other I/O, it could take up a few minutes. Therefore 64MB size is fairly good range for resync. - BARRIER_BUCKETS_NR There are BARRIER_BUCKETS_NR buckets in total, which is defined by, #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - 2) #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) this patch makes the bellowed members of struct r1conf from integer to array of integers, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int *nr_pending; + int *nr_waiting; + int *nr_queued; + int *barrier; number of the array elements is defined as BARRIER_BUCKETS_NR. For 4KB kernel space page size, (PAGE_SHIFT - 2) indecates there are 1024 I/O barrier buckets, and each array of integers occupies single memory page. 1024 means for a request which is smaller than the I/O barrier unit size has ~0.1% chance to wait for resync to pause, which is quite a small enough fraction. Also requesting single memory page is more friendly to kernel page allocator than larger memory size. - I/O barrier bucket is indexed by bio start sector If multiple I/O requests hit different I/O barrier units, they only need to compete I/O barrier with other I/Os which hit the same I/O barrier bucket index with each other. The index of a barrier bucket which a bio should look for is calculated by sector_to_idx() which is defined in raid1.h as an inline function, static inline int sector_to_idx(sector_t sector) { return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, BARRIER_BUCKETS_NR_BITS); } Here sector_nr is the start sector number of a bio. - Single bio won't go across boundary of a I/O barrier unit If a request goes across boundary of barrier unit, it will be split. A bio may be split in raid1_make_request() or raid1_sync_request(), if sectors returned by align_to_barrier_unit_end() is smaller than original bio size. Comparing to single sliding resync window, - Currently resync I/O grows linearly, therefore regular and resync I/O will conflict within a single barrier units. So the I/O behavior is similar to single sliding resync window. - But a barrier unit bucket is shared by all barrier units with identical barrier uinit index, the probability of conflict might be higher than single sliding resync window, in condition that writing I/Os always hit barrier units which have identical barrier bucket indexs with the resync I/Os. This is a very rare condition in real I/O work loads, I cannot imagine how it could happen in practice. - Therefore we can achieve a good enough low conflict rate with much simpler barrier algorithm and implementation. There are two changes should be noticed, - In raid1d(), I change the code to decrease conf->nr_pending[idx] into single loop, it looks like this, spin_lock_irqsave(&conf->device_lock, flags); conf->nr_queued[idx]--; spin_unlock_irqrestore(&conf->device_lock, flags); This change generates more spin lock operations, but in next patch of this patch set, it will be replaced by a single line code, atomic_dec(&conf->nr_queueud[idx]); So we don't need to worry about spin lock cost here. - Mainline raid1 code split original raid1_make_request() into raid1_read_request() and raid1_write_request(). If the original bio goes across an I/O barrier unit size, this bio will be split before calling raid1_read_request() or raid1_write_request(), this change the code logic more simple and clear. - In this patch wait_barrier() is moved from raid1_make_request() to raid1_write_request(). In raid_read_request(), original wait_barrier() is replaced by raid1_read_request(). The differnece is wait_read_barrier() only waits if array is frozen, using different barrier function in different code path makes the code more clean and easy to read. Changelog V4: - Add alloc_r1bio() to remove redundant r1bio memory allocation code. - Fix many typos in patch comments. - Use (PAGE_SHIFT - ilog2(sizeof(int))) to define BARRIER_BUCKETS_NR_BITS. V3: - Rebase the patch against latest upstream kernel code. - Many fixes by review comments from Neil, - Back to use pointers to replace arraries in struct r1conf - Remove total_barriers from struct r1conf - Add more patch comments to explain how/why the values of BARRIER_UNIT_SECTOR_SIZE and BARRIER_BUCKETS_NR are decided. - Use get_unqueued_pending() to replace get_all_pendings() and get_all_queued() - Increase bucket number from 512 to 1024 - Change code comments format by review from Shaohua. V2: - Use bio_split() to split the orignal bio if it goes across barrier unit bounday, to make the code more simple, by suggestion from Shaohua and Neil. - Use hash_long() to replace original linear hash, to avoid a possible confilict between resync I/O and sequential write I/O, by suggestion from Shaohua. - Add conf->total_barriers to record barrier depth, which is used to control number of parallel sync I/O barriers, by suggestion from Shaohua. - In V1 patch the bellowed barrier buckets related members in r1conf are allocated in memory page. To make the code more simple, V2 patch moves the memory space into struct r1conf, like this, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int nr_pending[BARRIER_BUCKETS_NR]; + int nr_waiting[BARRIER_BUCKETS_NR]; + int nr_queued[BARRIER_BUCKETS_NR]; + int barrier[BARRIER_BUCKETS_NR]; This change is by the suggestion from Shaohua. - Remove some inrelavent code comments, by suggestion from Guoqing. - Add a missing wait_barrier() before jumping to retry_write, in raid1_make_write_request(). V1: - Original RFC patch for comments Signed-off-by: Coly Li <colyli@suse.de> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:56 +00:00
*/
RAID1: avoid unnecessary spin locks in I/O barrier code When I run a parallel reading performan testing on a md raid1 device with two NVMe SSDs, I observe very bad throughput in supprise: by fio with 64KB block size, 40 seq read I/O jobs, 128 iodepth, overall throughput is only 2.7GB/s, this is around 50% of the idea performance number. The perf reports locking contention happens at allow_barrier() and wait_barrier() code, - 41.41% fio [kernel.kallsyms] [k] _raw_spin_lock_irqsave - _raw_spin_lock_irqsave + 89.92% allow_barrier + 9.34% __wake_up - 37.30% fio [kernel.kallsyms] [k] _raw_spin_lock_irq - _raw_spin_lock_irq - 100.00% wait_barrier The reason is, in these I/O barrier related functions, - raise_barrier() - lower_barrier() - wait_barrier() - allow_barrier() They always hold conf->resync_lock firstly, even there are only regular reading I/Os and no resync I/O at all. This is a huge performance penalty. The solution is a lockless-like algorithm in I/O barrier code, and only holding conf->resync_lock when it has to. The original idea is from Hannes Reinecke, and Neil Brown provides comments to improve it. I continue to work on it, and make the patch into current form. In the new simpler raid1 I/O barrier implementation, there are two wait barrier functions, - wait_barrier() Which calls _wait_barrier(), is used for regular write I/O. If there is resync I/O happening on the same I/O barrier bucket, or the whole array is frozen, task will wait until no barrier on same barrier bucket, or the whold array is unfreezed. - wait_read_barrier() Since regular read I/O won't interfere with resync I/O (read_balance() will make sure only uptodate data will be read out), it is unnecessary to wait for barrier in regular read I/Os, waiting in only necessary when the whole array is frozen. The operations on conf->nr_pending[idx], conf->nr_waiting[idx], conf-> barrier[idx] are very carefully designed in raise_barrier(), lower_barrier(), _wait_barrier() and wait_read_barrier(), in order to avoid unnecessary spin locks in these functions. Once conf-> nr_pengding[idx] is increased, a resync I/O with same barrier bucket index has to wait in raise_barrier(). Then in _wait_barrier() if no barrier raised in same barrier bucket index and array is not frozen, the regular I/O doesn't need to hold conf->resync_lock, it can just increase conf->nr_pending[idx], and return to its caller. wait_read_barrier() is very similar to _wait_barrier(), the only difference is it only waits when array is frozen. For heavy parallel reading I/Os, the lockless I/O barrier code almostly gets rid of all spin lock cost. This patch significantly improves raid1 reading peroformance. From my testing, a raid1 device built by two NVMe SSD, runs fio with 64KB blocksize, 40 seq read I/O jobs, 128 iodepth, overall throughput increases from 2.7GB/s to 4.6GB/s (+70%). Changelog V4: - Change conf->nr_queued[] to atomic_t. - Define BARRIER_BUCKETS_NR_BITS by (PAGE_SHIFT - ilog2(sizeof(atomic_t))) V3: - Add smp_mb__after_atomic() as Shaohua and Neil suggested. - Change conf->nr_queued[] from atomic_t to int. - Change conf->array_frozen from atomic_t back to int, and use READ_ONCE(conf->array_frozen) to check value of conf->array_frozen in _wait_barrier() and wait_read_barrier(). - In _wait_barrier() and wait_read_barrier(), add a call to wake_up(&conf->wait_barrier) after atomic_dec(&conf->nr_pending[idx]), to fix a deadlock between _wait_barrier()/wait_read_barrier and freeze_array(). V2: - Remove a spin_lock/unlock pair in raid1d(). - Add more code comments to explain why there is no racy when checking two atomic_t variables at same time. V1: - Original RFC patch for comments. Signed-off-by: Coly Li <colyli@suse.de> Cc: Shaohua Li <shli@fb.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:57 +00:00
#define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - ilog2(sizeof(atomic_t)))
RAID1: a new I/O barrier implementation to remove resync window 'Commit 79ef3a8aa1cb ("raid1: Rewrite the implementation of iobarrier.")' introduces a sliding resync window for raid1 I/O barrier, this idea limits I/O barriers to happen only inside a slidingresync window, for regular I/Os out of this resync window they don't need to wait for barrier any more. On large raid1 device, it helps a lot to improve parallel writing I/O throughput when there are background resync I/Os performing at same time. The idea of sliding resync widow is awesome, but code complexity is a challenge. Sliding resync window requires several variables to work collectively, this is complexed and very hard to make it work correctly. Just grep "Fixes: 79ef3a8aa1" in kernel git log, there are 8 more patches to fix the original resync window patch. This is not the end, any further related modification may easily introduce more regreassion. Therefore I decide to implement a much simpler raid1 I/O barrier, by removing resync window code, I believe life will be much easier. The brief idea of the simpler barrier is, - Do not maintain a global unique resync window - Use multiple hash buckets to reduce I/O barrier conflicts, regular I/O only has to wait for a resync I/O when both them have same barrier bucket index, vice versa. - I/O barrier can be reduced to an acceptable number if there are enough barrier buckets Here I explain how the barrier buckets are designed, - BARRIER_UNIT_SECTOR_SIZE The whole LBA address space of a raid1 device is divided into multiple barrier units, by the size of BARRIER_UNIT_SECTOR_SIZE. Bio requests won't go across border of barrier unit size, that means maximum bio size is BARRIER_UNIT_SECTOR_SIZE<<9 (64MB) in bytes. For random I/O 64MB is large enough for both read and write requests, for sequential I/O considering underlying block layer may merge them into larger requests, 64MB is still good enough. Neil also points out that for resync operation, "we want the resync to move from region to region fairly quickly so that the slowness caused by having to synchronize with the resync is averaged out over a fairly small time frame". For full speed resync, 64MB should take less then 1 second. When resync is competing with other I/O, it could take up a few minutes. Therefore 64MB size is fairly good range for resync. - BARRIER_BUCKETS_NR There are BARRIER_BUCKETS_NR buckets in total, which is defined by, #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - 2) #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) this patch makes the bellowed members of struct r1conf from integer to array of integers, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int *nr_pending; + int *nr_waiting; + int *nr_queued; + int *barrier; number of the array elements is defined as BARRIER_BUCKETS_NR. For 4KB kernel space page size, (PAGE_SHIFT - 2) indecates there are 1024 I/O barrier buckets, and each array of integers occupies single memory page. 1024 means for a request which is smaller than the I/O barrier unit size has ~0.1% chance to wait for resync to pause, which is quite a small enough fraction. Also requesting single memory page is more friendly to kernel page allocator than larger memory size. - I/O barrier bucket is indexed by bio start sector If multiple I/O requests hit different I/O barrier units, they only need to compete I/O barrier with other I/Os which hit the same I/O barrier bucket index with each other. The index of a barrier bucket which a bio should look for is calculated by sector_to_idx() which is defined in raid1.h as an inline function, static inline int sector_to_idx(sector_t sector) { return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, BARRIER_BUCKETS_NR_BITS); } Here sector_nr is the start sector number of a bio. - Single bio won't go across boundary of a I/O barrier unit If a request goes across boundary of barrier unit, it will be split. A bio may be split in raid1_make_request() or raid1_sync_request(), if sectors returned by align_to_barrier_unit_end() is smaller than original bio size. Comparing to single sliding resync window, - Currently resync I/O grows linearly, therefore regular and resync I/O will conflict within a single barrier units. So the I/O behavior is similar to single sliding resync window. - But a barrier unit bucket is shared by all barrier units with identical barrier uinit index, the probability of conflict might be higher than single sliding resync window, in condition that writing I/Os always hit barrier units which have identical barrier bucket indexs with the resync I/Os. This is a very rare condition in real I/O work loads, I cannot imagine how it could happen in practice. - Therefore we can achieve a good enough low conflict rate with much simpler barrier algorithm and implementation. There are two changes should be noticed, - In raid1d(), I change the code to decrease conf->nr_pending[idx] into single loop, it looks like this, spin_lock_irqsave(&conf->device_lock, flags); conf->nr_queued[idx]--; spin_unlock_irqrestore(&conf->device_lock, flags); This change generates more spin lock operations, but in next patch of this patch set, it will be replaced by a single line code, atomic_dec(&conf->nr_queueud[idx]); So we don't need to worry about spin lock cost here. - Mainline raid1 code split original raid1_make_request() into raid1_read_request() and raid1_write_request(). If the original bio goes across an I/O barrier unit size, this bio will be split before calling raid1_read_request() or raid1_write_request(), this change the code logic more simple and clear. - In this patch wait_barrier() is moved from raid1_make_request() to raid1_write_request(). In raid_read_request(), original wait_barrier() is replaced by raid1_read_request(). The differnece is wait_read_barrier() only waits if array is frozen, using different barrier function in different code path makes the code more clean and easy to read. Changelog V4: - Add alloc_r1bio() to remove redundant r1bio memory allocation code. - Fix many typos in patch comments. - Use (PAGE_SHIFT - ilog2(sizeof(int))) to define BARRIER_BUCKETS_NR_BITS. V3: - Rebase the patch against latest upstream kernel code. - Many fixes by review comments from Neil, - Back to use pointers to replace arraries in struct r1conf - Remove total_barriers from struct r1conf - Add more patch comments to explain how/why the values of BARRIER_UNIT_SECTOR_SIZE and BARRIER_BUCKETS_NR are decided. - Use get_unqueued_pending() to replace get_all_pendings() and get_all_queued() - Increase bucket number from 512 to 1024 - Change code comments format by review from Shaohua. V2: - Use bio_split() to split the orignal bio if it goes across barrier unit bounday, to make the code more simple, by suggestion from Shaohua and Neil. - Use hash_long() to replace original linear hash, to avoid a possible confilict between resync I/O and sequential write I/O, by suggestion from Shaohua. - Add conf->total_barriers to record barrier depth, which is used to control number of parallel sync I/O barriers, by suggestion from Shaohua. - In V1 patch the bellowed barrier buckets related members in r1conf are allocated in memory page. To make the code more simple, V2 patch moves the memory space into struct r1conf, like this, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int nr_pending[BARRIER_BUCKETS_NR]; + int nr_waiting[BARRIER_BUCKETS_NR]; + int nr_queued[BARRIER_BUCKETS_NR]; + int barrier[BARRIER_BUCKETS_NR]; This change is by the suggestion from Shaohua. - Remove some inrelavent code comments, by suggestion from Guoqing. - Add a missing wait_barrier() before jumping to retry_write, in raid1_make_write_request(). V1: - Original RFC patch for comments Signed-off-by: Coly Li <colyli@suse.de> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:56 +00:00
#define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS)
struct raid1_info {
struct md_rdev *rdev;
sector_t head_position;
/* When choose the best device for a read (read_balance())
* we try to keep sequential reads one the same device
*/
sector_t next_seq_sect;
sector_t seq_start;
};
/*
* memory pools need a pointer to the mddev, so they can force an unplug
* when memory is tight, and a count of the number of drives that the
* pool was allocated for, so they know how much to allocate and free.
* mddev->raid_disks cannot be used, as it can change while a pool is active
* These two datums are stored in a kmalloced struct.
* The 'raid_disks' here is twice the raid_disks in r1conf.
* This allows space for each 'real' device can have a replacement in the
* second half of the array.
*/
struct pool_info {
struct mddev *mddev;
int raid_disks;
};
struct r1conf {
struct mddev *mddev;
struct raid1_info *mirrors; /* twice 'raid_disks' to
* allow for replacements.
*/
int raid_disks;
spinlock_t device_lock;
/* list of 'struct r1bio' that need to be processed by raid1d,
* whether to retry a read, writeout a resync or recovery
* block, or anything else.
*/
struct list_head retry_list;
/* A separate list of r1bio which just need raid_end_bio_io called.
* This mustn't happen for writes which had any errors if the superblock
* needs to be written.
*/
struct list_head bio_end_io_list;
/* queue pending writes to be submitted on unplug */
struct bio_list pending_bio_list;
int pending_count;
/* for use when syncing mirrors:
* We don't allow both normal IO and resync/recovery IO at
* the same time - resync/recovery can only happen when there
* is no other IO. So when either is active, the other has to wait.
* See more details description in raid1.c near raise_barrier().
*/
wait_queue_head_t wait_barrier;
spinlock_t resync_lock;
atomic_t nr_sync_pending;
RAID1: avoid unnecessary spin locks in I/O barrier code When I run a parallel reading performan testing on a md raid1 device with two NVMe SSDs, I observe very bad throughput in supprise: by fio with 64KB block size, 40 seq read I/O jobs, 128 iodepth, overall throughput is only 2.7GB/s, this is around 50% of the idea performance number. The perf reports locking contention happens at allow_barrier() and wait_barrier() code, - 41.41% fio [kernel.kallsyms] [k] _raw_spin_lock_irqsave - _raw_spin_lock_irqsave + 89.92% allow_barrier + 9.34% __wake_up - 37.30% fio [kernel.kallsyms] [k] _raw_spin_lock_irq - _raw_spin_lock_irq - 100.00% wait_barrier The reason is, in these I/O barrier related functions, - raise_barrier() - lower_barrier() - wait_barrier() - allow_barrier() They always hold conf->resync_lock firstly, even there are only regular reading I/Os and no resync I/O at all. This is a huge performance penalty. The solution is a lockless-like algorithm in I/O barrier code, and only holding conf->resync_lock when it has to. The original idea is from Hannes Reinecke, and Neil Brown provides comments to improve it. I continue to work on it, and make the patch into current form. In the new simpler raid1 I/O barrier implementation, there are two wait barrier functions, - wait_barrier() Which calls _wait_barrier(), is used for regular write I/O. If there is resync I/O happening on the same I/O barrier bucket, or the whole array is frozen, task will wait until no barrier on same barrier bucket, or the whold array is unfreezed. - wait_read_barrier() Since regular read I/O won't interfere with resync I/O (read_balance() will make sure only uptodate data will be read out), it is unnecessary to wait for barrier in regular read I/Os, waiting in only necessary when the whole array is frozen. The operations on conf->nr_pending[idx], conf->nr_waiting[idx], conf-> barrier[idx] are very carefully designed in raise_barrier(), lower_barrier(), _wait_barrier() and wait_read_barrier(), in order to avoid unnecessary spin locks in these functions. Once conf-> nr_pengding[idx] is increased, a resync I/O with same barrier bucket index has to wait in raise_barrier(). Then in _wait_barrier() if no barrier raised in same barrier bucket index and array is not frozen, the regular I/O doesn't need to hold conf->resync_lock, it can just increase conf->nr_pending[idx], and return to its caller. wait_read_barrier() is very similar to _wait_barrier(), the only difference is it only waits when array is frozen. For heavy parallel reading I/Os, the lockless I/O barrier code almostly gets rid of all spin lock cost. This patch significantly improves raid1 reading peroformance. From my testing, a raid1 device built by two NVMe SSD, runs fio with 64KB blocksize, 40 seq read I/O jobs, 128 iodepth, overall throughput increases from 2.7GB/s to 4.6GB/s (+70%). Changelog V4: - Change conf->nr_queued[] to atomic_t. - Define BARRIER_BUCKETS_NR_BITS by (PAGE_SHIFT - ilog2(sizeof(atomic_t))) V3: - Add smp_mb__after_atomic() as Shaohua and Neil suggested. - Change conf->nr_queued[] from atomic_t to int. - Change conf->array_frozen from atomic_t back to int, and use READ_ONCE(conf->array_frozen) to check value of conf->array_frozen in _wait_barrier() and wait_read_barrier(). - In _wait_barrier() and wait_read_barrier(), add a call to wake_up(&conf->wait_barrier) after atomic_dec(&conf->nr_pending[idx]), to fix a deadlock between _wait_barrier()/wait_read_barrier and freeze_array(). V2: - Remove a spin_lock/unlock pair in raid1d(). - Add more code comments to explain why there is no racy when checking two atomic_t variables at same time. V1: - Original RFC patch for comments. Signed-off-by: Coly Li <colyli@suse.de> Cc: Shaohua Li <shli@fb.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:57 +00:00
atomic_t *nr_pending;
atomic_t *nr_waiting;
atomic_t *nr_queued;
atomic_t *barrier;
int array_frozen;
/* Set to 1 if a full sync is needed, (fresh device added).
* Cleared when a sync completes.
*/
int fullsync;
/* When the same as mddev->recovery_disabled we don't allow
* recovery to be attempted as we expect a read error.
*/
int recovery_disabled;
/* poolinfo contains information about the content of the
* mempools - it changes when the array grows or shrinks
*/
struct pool_info *poolinfo;
mempool_t *r1bio_pool;
mempool_t *r1buf_pool;
struct bio_set *bio_split;
/* temporary buffer to synchronous IO when attempting to repair
* a read error.
*/
struct page *tmppage;
/* When taking over an array from a different personality, we store
* the new thread here until we fully activate the array.
*/
struct md_thread *thread;
/* Keep track of cluster resync window to send to other
* nodes.
*/
sector_t cluster_sync_low;
sector_t cluster_sync_high;
};
/*
* this is our 'private' RAID1 bio.
*
* it contains information about what kind of IO operations were started
* for this RAID1 operation, and about their status:
*/
struct r1bio {
atomic_t remaining; /* 'have we finished' count,
* used from IRQ handlers
*/
atomic_t behind_remaining; /* number of write-behind ios remaining
* in this BehindIO request
*/
sector_t sector;
int sectors;
unsigned long state;
struct mddev *mddev;
/*
* original bio going to /dev/mdx
*/
struct bio *master_bio;
/*
* if the IO is in READ direction, then this is where we read
*/
int read_disk;
struct list_head retry_list;
/*
* When R1BIO_BehindIO is set, we store pages for write behind
* in behind_master_bio.
*/
struct bio *behind_master_bio;
/*
* if the IO is in WRITE direction, then multiple bios are used.
* We choose the number when they are allocated.
*/
struct bio *bios[0];
/* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/
};
/* bits for r1bio.state */
enum r1bio_state {
R1BIO_Uptodate,
R1BIO_IsSync,
R1BIO_Degraded,
R1BIO_BehindIO,
/* Set ReadError on bios that experience a readerror so that
* raid1d knows what to do with them.
*/
R1BIO_ReadError,
/* For write-behind requests, we call bi_end_io when
* the last non-write-behind device completes, providing
* any write was successful. Otherwise we call when
* any write-behind write succeeds, otherwise we call
* with failure when last write completes (and all failed).
* Record that bi_end_io was called with this flag...
*/
R1BIO_Returned,
/* If a write for this request means we can clear some
* known-bad-block records, we set this flag
*/
R1BIO_MadeGood,
R1BIO_WriteError,
md/raid1: add failfast handling for reads. If a device is marked FailFast and it is not the only device we can read from, we mark the bio with REQ_FAILFAST_* flags. If this does fail, we don't try read repair but just allow failure. If it was the last device it doesn't fail of course, so the retry happens on the same device - this time without FAILFAST. A subsequent failure will not retry but will just pass up the error. During resync we may use FAILFAST requests and on a failure we will simply use the other device(s). During recovery we will only use FAILFAST in the unusual case were there are multiple places to read from - i.e. if there are > 2 devices. If we get a failure we will fail the device and complete the resync/recovery with remaining devices. The new R1BIO_FailFast flag is set on read reqest to suggest the a FAILFAST request might be acceptable. The rdev needs to have FailFast set as well for the read to actually use REQ_FAILFAST_*. We need to know there are at least two working devices before we can set R1BIO_FailFast, so we mustn't stop looking at the first device we find. So the "min_pending == 0" handling to not exit early, but too always choose the best_pending_disk if min_pending == 0. The spinlocked region in raid1_error() in enlarged to ensure that if two bios, reading from two different devices, fail at the same time, then there is no risk that both devices will be marked faulty, leaving zero "In_sync" devices. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Shaohua Li <shli@fb.com>
2016-11-18 05:16:12 +00:00
R1BIO_FailFast,
};
RAID1: a new I/O barrier implementation to remove resync window 'Commit 79ef3a8aa1cb ("raid1: Rewrite the implementation of iobarrier.")' introduces a sliding resync window for raid1 I/O barrier, this idea limits I/O barriers to happen only inside a slidingresync window, for regular I/Os out of this resync window they don't need to wait for barrier any more. On large raid1 device, it helps a lot to improve parallel writing I/O throughput when there are background resync I/Os performing at same time. The idea of sliding resync widow is awesome, but code complexity is a challenge. Sliding resync window requires several variables to work collectively, this is complexed and very hard to make it work correctly. Just grep "Fixes: 79ef3a8aa1" in kernel git log, there are 8 more patches to fix the original resync window patch. This is not the end, any further related modification may easily introduce more regreassion. Therefore I decide to implement a much simpler raid1 I/O barrier, by removing resync window code, I believe life will be much easier. The brief idea of the simpler barrier is, - Do not maintain a global unique resync window - Use multiple hash buckets to reduce I/O barrier conflicts, regular I/O only has to wait for a resync I/O when both them have same barrier bucket index, vice versa. - I/O barrier can be reduced to an acceptable number if there are enough barrier buckets Here I explain how the barrier buckets are designed, - BARRIER_UNIT_SECTOR_SIZE The whole LBA address space of a raid1 device is divided into multiple barrier units, by the size of BARRIER_UNIT_SECTOR_SIZE. Bio requests won't go across border of barrier unit size, that means maximum bio size is BARRIER_UNIT_SECTOR_SIZE<<9 (64MB) in bytes. For random I/O 64MB is large enough for both read and write requests, for sequential I/O considering underlying block layer may merge them into larger requests, 64MB is still good enough. Neil also points out that for resync operation, "we want the resync to move from region to region fairly quickly so that the slowness caused by having to synchronize with the resync is averaged out over a fairly small time frame". For full speed resync, 64MB should take less then 1 second. When resync is competing with other I/O, it could take up a few minutes. Therefore 64MB size is fairly good range for resync. - BARRIER_BUCKETS_NR There are BARRIER_BUCKETS_NR buckets in total, which is defined by, #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - 2) #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) this patch makes the bellowed members of struct r1conf from integer to array of integers, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int *nr_pending; + int *nr_waiting; + int *nr_queued; + int *barrier; number of the array elements is defined as BARRIER_BUCKETS_NR. For 4KB kernel space page size, (PAGE_SHIFT - 2) indecates there are 1024 I/O barrier buckets, and each array of integers occupies single memory page. 1024 means for a request which is smaller than the I/O barrier unit size has ~0.1% chance to wait for resync to pause, which is quite a small enough fraction. Also requesting single memory page is more friendly to kernel page allocator than larger memory size. - I/O barrier bucket is indexed by bio start sector If multiple I/O requests hit different I/O barrier units, they only need to compete I/O barrier with other I/Os which hit the same I/O barrier bucket index with each other. The index of a barrier bucket which a bio should look for is calculated by sector_to_idx() which is defined in raid1.h as an inline function, static inline int sector_to_idx(sector_t sector) { return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, BARRIER_BUCKETS_NR_BITS); } Here sector_nr is the start sector number of a bio. - Single bio won't go across boundary of a I/O barrier unit If a request goes across boundary of barrier unit, it will be split. A bio may be split in raid1_make_request() or raid1_sync_request(), if sectors returned by align_to_barrier_unit_end() is smaller than original bio size. Comparing to single sliding resync window, - Currently resync I/O grows linearly, therefore regular and resync I/O will conflict within a single barrier units. So the I/O behavior is similar to single sliding resync window. - But a barrier unit bucket is shared by all barrier units with identical barrier uinit index, the probability of conflict might be higher than single sliding resync window, in condition that writing I/Os always hit barrier units which have identical barrier bucket indexs with the resync I/Os. This is a very rare condition in real I/O work loads, I cannot imagine how it could happen in practice. - Therefore we can achieve a good enough low conflict rate with much simpler barrier algorithm and implementation. There are two changes should be noticed, - In raid1d(), I change the code to decrease conf->nr_pending[idx] into single loop, it looks like this, spin_lock_irqsave(&conf->device_lock, flags); conf->nr_queued[idx]--; spin_unlock_irqrestore(&conf->device_lock, flags); This change generates more spin lock operations, but in next patch of this patch set, it will be replaced by a single line code, atomic_dec(&conf->nr_queueud[idx]); So we don't need to worry about spin lock cost here. - Mainline raid1 code split original raid1_make_request() into raid1_read_request() and raid1_write_request(). If the original bio goes across an I/O barrier unit size, this bio will be split before calling raid1_read_request() or raid1_write_request(), this change the code logic more simple and clear. - In this patch wait_barrier() is moved from raid1_make_request() to raid1_write_request(). In raid_read_request(), original wait_barrier() is replaced by raid1_read_request(). The differnece is wait_read_barrier() only waits if array is frozen, using different barrier function in different code path makes the code more clean and easy to read. Changelog V4: - Add alloc_r1bio() to remove redundant r1bio memory allocation code. - Fix many typos in patch comments. - Use (PAGE_SHIFT - ilog2(sizeof(int))) to define BARRIER_BUCKETS_NR_BITS. V3: - Rebase the patch against latest upstream kernel code. - Many fixes by review comments from Neil, - Back to use pointers to replace arraries in struct r1conf - Remove total_barriers from struct r1conf - Add more patch comments to explain how/why the values of BARRIER_UNIT_SECTOR_SIZE and BARRIER_BUCKETS_NR are decided. - Use get_unqueued_pending() to replace get_all_pendings() and get_all_queued() - Increase bucket number from 512 to 1024 - Change code comments format by review from Shaohua. V2: - Use bio_split() to split the orignal bio if it goes across barrier unit bounday, to make the code more simple, by suggestion from Shaohua and Neil. - Use hash_long() to replace original linear hash, to avoid a possible confilict between resync I/O and sequential write I/O, by suggestion from Shaohua. - Add conf->total_barriers to record barrier depth, which is used to control number of parallel sync I/O barriers, by suggestion from Shaohua. - In V1 patch the bellowed barrier buckets related members in r1conf are allocated in memory page. To make the code more simple, V2 patch moves the memory space into struct r1conf, like this, - int nr_pending; - int nr_waiting; - int nr_queued; - int barrier; + int nr_pending[BARRIER_BUCKETS_NR]; + int nr_waiting[BARRIER_BUCKETS_NR]; + int nr_queued[BARRIER_BUCKETS_NR]; + int barrier[BARRIER_BUCKETS_NR]; This change is by the suggestion from Shaohua. - Remove some inrelavent code comments, by suggestion from Guoqing. - Add a missing wait_barrier() before jumping to retry_write, in raid1_make_write_request(). V1: - Original RFC patch for comments Signed-off-by: Coly Li <colyli@suse.de> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Guoqing Jiang <gqjiang@suse.com> Reviewed-by: Neil Brown <neilb@suse.de> Signed-off-by: Shaohua Li <shli@fb.com>
2017-02-17 19:05:56 +00:00
static inline int sector_to_idx(sector_t sector)
{
return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS,
BARRIER_BUCKETS_NR_BITS);
}
#endif