linux/drivers/md/dm-delay.c

440 lines
9.5 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright (C) 2005-2007 Red Hat GmbH
*
* A target that delays reads and/or writes and can send
* them to different devices.
*
* This file is released under the GPL.
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/slab.h>
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
#include <linux/kthread.h>
#include <linux/device-mapper.h>
#define DM_MSG_PREFIX "delay"
struct delay_class {
struct dm_dev *dev;
sector_t start;
unsigned int delay;
unsigned int ops;
};
struct delay_c {
struct timer_list delay_timer;
struct mutex timer_lock;
struct workqueue_struct *kdelayd_wq;
struct work_struct flush_expired_bios;
struct list_head delayed_bios;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
struct task_struct *worker;
bool may_delay;
struct delay_class read;
struct delay_class write;
struct delay_class flush;
int argc;
};
struct dm_delay_info {
struct delay_c *context;
struct delay_class *class;
struct list_head list;
unsigned long expires;
};
static DEFINE_MUTEX(delayed_bios_lock);
static void handle_delayed_timer(struct timer_list *t)
{
struct delay_c *dc = from_timer(dc, t, delay_timer);
queue_work(dc->kdelayd_wq, &dc->flush_expired_bios);
}
static void queue_timeout(struct delay_c *dc, unsigned long expires)
{
mutex_lock(&dc->timer_lock);
if (!timer_pending(&dc->delay_timer) || expires < dc->delay_timer.expires)
mod_timer(&dc->delay_timer, expires);
mutex_unlock(&dc->timer_lock);
}
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
static inline bool delay_is_fast(struct delay_c *dc)
{
return !!dc->worker;
}
static void flush_bios(struct bio *bio)
{
struct bio *n;
while (bio) {
n = bio->bi_next;
bio->bi_next = NULL;
dm_submit_bio_remap(bio, NULL);
bio = n;
}
}
static void flush_delayed_bios(struct delay_c *dc, bool flush_all)
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
{
struct dm_delay_info *delayed, *next;
struct bio_list flush_bio_list;
unsigned long next_expires = 0;
bool start_timer = false;
bio_list_init(&flush_bio_list);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
mutex_lock(&delayed_bios_lock);
list_for_each_entry_safe(delayed, next, &dc->delayed_bios, list) {
cond_resched();
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (flush_all || time_after_eq(jiffies, delayed->expires)) {
struct bio *bio = dm_bio_from_per_bio_data(delayed,
sizeof(struct dm_delay_info));
list_del(&delayed->list);
bio_list_add(&flush_bio_list, bio);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
delayed->class->ops--;
continue;
}
if (!delay_is_fast(dc)) {
if (!start_timer) {
start_timer = true;
next_expires = delayed->expires;
} else {
next_expires = min(next_expires, delayed->expires);
}
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
}
}
mutex_unlock(&delayed_bios_lock);
if (start_timer)
queue_timeout(dc, next_expires);
flush_bios(bio_list_get(&flush_bio_list));
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
}
static int flush_worker_fn(void *data)
{
struct delay_c *dc = data;
while (!kthread_should_stop()) {
flush_delayed_bios(dc, false);
mutex_lock(&delayed_bios_lock);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (unlikely(list_empty(&dc->delayed_bios))) {
set_current_state(TASK_INTERRUPTIBLE);
mutex_unlock(&delayed_bios_lock);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
schedule();
} else {
mutex_unlock(&delayed_bios_lock);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
cond_resched();
}
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
}
return 0;
}
static void flush_expired_bios(struct work_struct *work)
{
struct delay_c *dc;
dc = container_of(work, struct delay_c, flush_expired_bios);
flush_delayed_bios(dc, false);
}
static void delay_dtr(struct dm_target *ti)
{
struct delay_c *dc = ti->private;
if (dc->kdelayd_wq)
destroy_workqueue(dc->kdelayd_wq);
if (dc->read.dev)
dm_put_device(ti, dc->read.dev);
if (dc->write.dev)
dm_put_device(ti, dc->write.dev);
if (dc->flush.dev)
dm_put_device(ti, dc->flush.dev);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (dc->worker)
kthread_stop(dc->worker);
mutex_destroy(&dc->timer_lock);
kfree(dc);
}
static int delay_class_ctr(struct dm_target *ti, struct delay_class *c, char **argv)
{
int ret;
unsigned long long tmpll;
char dummy;
if (sscanf(argv[1], "%llu%c", &tmpll, &dummy) != 1 || tmpll != (sector_t)tmpll) {
ti->error = "Invalid device sector";
return -EINVAL;
}
c->start = tmpll;
if (sscanf(argv[2], "%u%c", &c->delay, &dummy) != 1) {
ti->error = "Invalid delay";
return -EINVAL;
}
ret = dm_get_device(ti, argv[0], dm_table_get_mode(ti->table), &c->dev);
if (ret) {
ti->error = "Device lookup failed";
return ret;
}
return 0;
}
/*
* Mapping parameters:
* <device> <offset> <delay> [<write_device> <write_offset> <write_delay>]
*
* With separate write parameters, the first set is only used for reads.
* Offsets are specified in sectors.
* Delays are specified in milliseconds.
*/
static int delay_ctr(struct dm_target *ti, unsigned int argc, char **argv)
{
struct delay_c *dc;
int ret;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
unsigned int max_delay;
if (argc != 3 && argc != 6 && argc != 9) {
ti->error = "Requires exactly 3, 6 or 9 arguments";
return -EINVAL;
}
dc = kzalloc(sizeof(*dc), GFP_KERNEL);
if (!dc) {
ti->error = "Cannot allocate context";
return -ENOMEM;
}
ti->private = dc;
INIT_LIST_HEAD(&dc->delayed_bios);
mutex_init(&dc->timer_lock);
dc->may_delay = true;
dc->argc = argc;
ret = delay_class_ctr(ti, &dc->read, argv);
if (ret)
goto bad;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
max_delay = dc->read.delay;
if (argc == 3) {
ret = delay_class_ctr(ti, &dc->write, argv);
if (ret)
goto bad;
ret = delay_class_ctr(ti, &dc->flush, argv);
if (ret)
goto bad;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
max_delay = max(max_delay, dc->write.delay);
max_delay = max(max_delay, dc->flush.delay);
goto out;
}
ret = delay_class_ctr(ti, &dc->write, argv + 3);
if (ret)
goto bad;
if (argc == 6) {
ret = delay_class_ctr(ti, &dc->flush, argv + 3);
if (ret)
goto bad;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
max_delay = max(max_delay, dc->flush.delay);
goto out;
}
ret = delay_class_ctr(ti, &dc->flush, argv + 6);
if (ret)
goto bad;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
max_delay = max(max_delay, dc->flush.delay);
out:
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (max_delay < 50) {
/*
* In case of small requested delays, use kthread instead of
* timers and workqueue to achieve better latency.
*/
dc->worker = kthread_create(&flush_worker_fn, dc,
"dm-delay-flush-worker");
if (IS_ERR(dc->worker)) {
ret = PTR_ERR(dc->worker);
dc->worker = NULL;
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
goto bad;
}
} else {
timer_setup(&dc->delay_timer, handle_delayed_timer, 0);
INIT_WORK(&dc->flush_expired_bios, flush_expired_bios);
dc->kdelayd_wq = alloc_workqueue("kdelayd", WQ_MEM_RECLAIM, 0);
if (!dc->kdelayd_wq) {
ret = -EINVAL;
DMERR("Couldn't start kdelayd");
goto bad;
}
}
ti->num_flush_bios = 1;
ti->num_discard_bios = 1;
ti->accounts_remapped_io = true;
ti->per_io_data_size = sizeof(struct dm_delay_info);
return 0;
bad:
delay_dtr(ti);
return ret;
}
static int delay_bio(struct delay_c *dc, struct delay_class *c, struct bio *bio)
{
struct dm_delay_info *delayed;
unsigned long expires = 0;
if (!c->delay)
return DM_MAPIO_REMAPPED;
delayed = dm_per_bio_data(bio, sizeof(struct dm_delay_info));
delayed->context = dc;
delayed->expires = expires = jiffies + msecs_to_jiffies(c->delay);
mutex_lock(&delayed_bios_lock);
if (unlikely(!dc->may_delay)) {
mutex_unlock(&delayed_bios_lock);
return DM_MAPIO_REMAPPED;
}
c->ops++;
list_add_tail(&delayed->list, &dc->delayed_bios);
mutex_unlock(&delayed_bios_lock);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (delay_is_fast(dc))
wake_up_process(dc->worker);
else
queue_timeout(dc, expires);
return DM_MAPIO_SUBMITTED;
}
static void delay_presuspend(struct dm_target *ti)
{
struct delay_c *dc = ti->private;
mutex_lock(&delayed_bios_lock);
dc->may_delay = false;
mutex_unlock(&delayed_bios_lock);
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
if (!delay_is_fast(dc))
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
del_timer_sync(&dc->delay_timer);
flush_delayed_bios(dc, true);
}
static void delay_resume(struct dm_target *ti)
{
struct delay_c *dc = ti->private;
dc->may_delay = true;
}
static int delay_map(struct dm_target *ti, struct bio *bio)
{
struct delay_c *dc = ti->private;
struct delay_class *c;
struct dm_delay_info *delayed = dm_per_bio_data(bio, sizeof(struct dm_delay_info));
if (bio_data_dir(bio) == WRITE) {
if (unlikely(bio->bi_opf & REQ_PREFLUSH))
c = &dc->flush;
else
c = &dc->write;
} else {
c = &dc->read;
}
delayed->class = c;
bio_set_dev(bio, c->dev->bdev);
bio->bi_iter.bi_sector = c->start + dm_target_offset(ti, bio->bi_iter.bi_sector);
return delay_bio(dc, c, bio);
}
#define DMEMIT_DELAY_CLASS(c) \
DMEMIT("%s %llu %u", (c)->dev->name, (unsigned long long)(c)->start, (c)->delay)
dm: fix truncated status strings Avoid returning a truncated table or status string instead of setting the DM_BUFFER_FULL_FLAG when the last target of a table fills the buffer. When processing a table or status request, the function retrieve_status calls ti->type->status. If ti->type->status returns non-zero, retrieve_status assumes that the buffer overflowed and sets DM_BUFFER_FULL_FLAG. However, targets don't return non-zero values from their status method on overflow. Most targets returns always zero. If a buffer overflow happens in a target that is not the last in the table, it gets noticed during the next iteration of the loop in retrieve_status; but if a buffer overflow happens in the last target, it goes unnoticed and erroneously truncated data is returned. In the current code, the targets behave in the following way: * dm-crypt returns -ENOMEM if there is not enough space to store the key, but it returns 0 on all other overflows. * dm-thin returns errors from the status method if a disk error happened. This is incorrect because retrieve_status doesn't check the error code, it assumes that all non-zero values mean buffer overflow. * all the other targets always return 0. This patch changes the ti->type->status function to return void (because most targets don't use the return code). Overflow is detected in retrieve_status: if the status method fills up the remaining space completely, it is assumed that buffer overflow happened. Cc: stable@vger.kernel.org Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2013-03-02 06:45:44 +08:00
static void delay_status(struct dm_target *ti, status_type_t type,
unsigned int status_flags, char *result, unsigned int maxlen)
{
struct delay_c *dc = ti->private;
int sz = 0;
switch (type) {
case STATUSTYPE_INFO:
DMEMIT("%u %u %u", dc->read.ops, dc->write.ops, dc->flush.ops);
break;
case STATUSTYPE_TABLE:
DMEMIT_DELAY_CLASS(&dc->read);
if (dc->argc >= 6) {
DMEMIT(" ");
DMEMIT_DELAY_CLASS(&dc->write);
}
if (dc->argc >= 9) {
DMEMIT(" ");
DMEMIT_DELAY_CLASS(&dc->flush);
}
break;
2021-07-13 08:49:03 +08:00
case STATUSTYPE_IMA:
*result = '\0';
break;
}
}
static int delay_iterate_devices(struct dm_target *ti,
iterate_devices_callout_fn fn, void *data)
{
struct delay_c *dc = ti->private;
int ret = 0;
ret = fn(ti, dc->read.dev, dc->read.start, ti->len, data);
if (ret)
goto out;
ret = fn(ti, dc->write.dev, dc->write.start, ti->len, data);
if (ret)
goto out;
ret = fn(ti, dc->flush.dev, dc->flush.start, ti->len, data);
if (ret)
goto out;
out:
return ret;
}
static struct target_type delay_target = {
.name = "delay",
dm delay: for short delays, use kthread instead of timers and wq DM delay's current design of using timers and wq to realize the delays is insufficient for delays below ~50ms. This commit enhances the design to use a kthread to flush the expired delays, trading some CPU time (in some cases) for better delay accuracy and delays closer to what the user requested for smaller delays. The new design is chosen as long as all the delays are below 50ms. Since bios can't be completed in interrupt context using a kthread is probably the most reasonable way to approach this. Testing with echo "0 2097152 zero" | dmsetup create dm-zeros for i in $(seq 0 20); do echo "0 2097152 delay /dev/mapper/dm-zeros 0 $i" | dmsetup create dm-delay-${i}ms; done Some performance numbers for comparison, on beaglebone black (single core) CONFIG_HZ_1000=y: fio --name=1msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 250 IOPS Kthread: 500 IOPS fio --name=10msread --rw=randread --bs=4k --runtime=60 --time_based \ --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 45 IOPS Kthread: 50 IOPS fio --name=1mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-1ms Theoretical maximum: 1000 IOPS Previous: 498 IOPS Kthread: 1000 IOPS fio --name=10mswrite --rw=randwrite --direct=1 --bs=4k --runtime=60 \ --time_based --filename=/dev/mapper/dm-delay-10ms Theoretical maximum: 100 IOPS Previous: 90 IOPS Kthread: 100 IOPS (This one is just to prove the new design isn't impacting throughput, not really about delays): fio --name=10mswriteasync --rw=randwrite --direct=1 --bs=4k \ --runtime=60 --time_based --filename=/dev/mapper/dm-delay-10ms \ --numjobs=32 --iodepth=64 --ioengine=libaio --group_reporting Previous: 13.3k IOPS Kthread: 13.3k IOPS Signed-off-by: Christian Loehle <christian.loehle@arm.com> [Harshit: kthread_create error handling fix in delay_ctr] Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org>
2023-10-20 19:46:05 +08:00
.version = {1, 4, 0},
.features = DM_TARGET_PASSES_INTEGRITY,
.module = THIS_MODULE,
.ctr = delay_ctr,
.dtr = delay_dtr,
.map = delay_map,
.presuspend = delay_presuspend,
.resume = delay_resume,
.status = delay_status,
.iterate_devices = delay_iterate_devices,
};
module_dm(delay);
MODULE_DESCRIPTION(DM_NAME " delay target");
MODULE_AUTHOR("Heinz Mauelshagen <mauelshagen@redhat.com>");
MODULE_LICENSE("GPL");