Skip to contentSkip to conversation

bcache: allow user to set QUEUE_FLAG_NOWAIT

PATCH In review Chaitanya Kulkarni 3 participants 4 messages May 12, 2023 – May 13, 2023 bcache (block layer cache)

lore
Share & tools
lore
#0 lore raw

cover bcache: allow user to set QUEUE_FLAG_NOWAIT

Allow user to set the QUEUE_FLAG_NOWAIT optionally using module parameter to retain the default behaviour. Also, update respective allocation flags in the write path. Following are the performance numbers with io_uring fio engine for random read, note that device has been populated fully with randwrite workload before taking these numbers :-

* linux-block (for-next) # grep IOPS bc-*fio | column -t

nowait-off-1.fio:  read:  IOPS=482k,  BW=1885MiB/s
nowait-off-2.fio:  read:  IOPS=484k,  BW=1889MiB/s
nowait-off-3.fio:  read:  IOPS=483k,  BW=1886MiB/s
nowait-on-1.fio:   read:  IOPS=544k,  BW=2125MiB/s
nowait-on-2.fio:   read:  IOPS=547k,  BW=2137MiB/s
nowait-on-3.fio:   read:  IOPS=546k,  BW=2132MiB/s

* linux-block (for-next) # grep slat bc-*fio | column -t

nowait-off-1.fio: slat (nsec): min=430, max=5488.5k, avg=2797.52 nowait-off-2.fio: slat (nsec): min=431, max=8252.4k, avg=2805.33 nowait-off-3.fio: slat (nsec): min=431, max=6846.6k, avg=2814.57

nowait-on-1.fio:  slat (usec):  min=2,   max=39086,   avg=87.48
nowait-on-2.fio:  slat (usec):  min=3,   max=39519,   avg=86.98
nowait-on-3.fio:  slat (usec):  min=3,   max=38880,   avg=87.17

* linux-block (for-next) # grep cpu bc-*fio | column -t

nowait-off-1.fio:  cpu  :  usr=2.77%,  sys=6.57%,   ctx=22015526
nowait-off-2.fio:  cpu  :  usr=2.75%,  sys=6.59%,   ctx=22003700
nowait-off-3.fio:  cpu  :  usr=2.81%,  sys=6.57%,   ctx=21938309
nowait-on-1.fio:   cpu  :  usr=1.08%,  sys=78.39%,  ctx=2744092
nowait-on-2.fio:   cpu  :  usr=1.10%,  sys=79.76%,  ctx=2537466
nowait-on-3.fio:   cpu  :  usr=1.10%,  sys=79.88%,  ctx=2528092

Please let me know if further testing is needed I've ran fio verification job in order to make verify these changes.

-ck

Chaitanya Kulkarni (1):
bcache: allow user to set QUEUE_FLAG_NOWAIT

 drivers/md/bcache/request.c | 3 ++-
 drivers/md/bcache/super.c   | 6 ++++++
 2 files changed, 8 insertions(+), 1 deletion(-)
linux-block (for-next) # sh test-bcache.sh 
+ git log -1
commit de812ff5f3c9819c83d6eaaed076a031a4b7be4a (HEAD -> for-next)
Author: Chaitanya Kulkarni <[email protected]>
Date:   Fri May 12 02:08:19 2023 -0700

bcache: allow user to set QUEUE_FLAG_NOWAIT

Allow user to set the QUEUE_FLAG_NOWAIT optionally using module parameter to retain the default behaviour. Also, update respective allocation flags in the write path. Following are the performance numbers with io_uring fio engine for random read, note that device has been populated fully with randwrite workload before taking these numbers :-
+ rmmod bcache.ko
rmmod: ERROR: Module bcache is not currently loaded
+ rmmod null_blk
rmmod: ERROR: Module null_blk is not currently loaded
+ makej M=drivers/md/bcache
+ modprobe null_blk queue_mode=2 nr_devices=2 memory_backed=1 gb=1
+ insmod drivers/md/bcache/bcache.ko
+ test_bcache default-nowait-off
+ bcache make -B /dev/nullb0 -C /dev/nullb1
Name /dev/nullb1
Label
Type cache
UUID: 36da1586-91a1-4629-a639-ed07d4a8e4fa
Set UUID: 614c08a8-975f-4eb4-a047-b589e586ffeb
version: 0
nbuckets: 2048
block_size_in_sectors: 1
bucket_size_in_sectors: 1024
nr_in_set: 1
nr_this_dev: 0
first_bucket: 1

                                ...

Name /dev/nullb0
Label
Type data
UUID: 1828367c-7859-4533-830e-b9594bc7ada5
Set UUID: 614c08a8-975f-4eb4-a047-b589e586ffeb
version: 1
block_size_in_sectors: 1
data_offset_in_sectors: 16

+ sleep 1
+ echo /dev/nullb0
+ echo /dev/nullb1
+ dmesg -c

[ 2428.580341] bcache: bcache_device_free() bcache0 stopped
[ 2428.581555] bcache: cache_set_free() Cache set 5eecd8b5-12f9-4863-9c31-a7cf4c98e080 unregistered
[ 2434.003724] null_blk: disk nullb0 created
[ 2434.004672] null_blk: disk nullb1 created
[ 2434.004674] null_blk: module loaded
[ 2435.037634] bcache: register_bdev() registered backing device nullb0
[ 2435.038927] bcache: run_cache_set() invalidating existing data
[ 2435.044138] bcache: bch_cached_dev_run() cached dev nullb0 is running already
[ 2435.044149] bcache: bch_cached_dev_attach() Caching nullb0 as bcache0 on set 614c08a8-975f-4eb4-a047-b589e586ffeb
[ 2435.044162] bcache: register_cache() registered cache device nullb1
+ lsblk
NAME      MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda         8:0    0   50G  0 disk 
├─sda1      8:1    0    1G  0 part /boot
└─sda2      8:2    0   49G  0 part /home
sdb         8:16   0  100G  0 disk /mnt/data
sr0        11:0    1 1024M  0 rom  
nullb0    251:0    0    1G  0 disk 
└─bcache0 250:0    0 1024M  0 disk 
nullb1    251:1    0    1G  0 disk 
└─bcache0 250:0    0 1024M  0 disk 
vda       252:0    0    5G  0 disk 
nvme0n1   259:0    0    1G  0 disk 

+ sleep 1
+ fio fio/verify.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0
write-and-verify: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=16 fio-3.34
Starting 1 process
verify: bad header numberio 63917, wanted 64003 at file /dev/bcache0 offset 701530112, length 4096 (requested block: offset=701530112, length=4096)
fio: pid=9319, err=84/file:io_u.c:2203, func=io_u_queued_complete, error=Invalid or incomplete multibyte or wide character

write-and-verify: (groupid=0, jobs=1): err=84 (file:io_u.c:2203, func=io_u_queued_complete, error=Invalid or incomplete multibyte or wide character): pid=9319: Fri May 12 02:35:17 2023
read: IOPS=237k, BW=925MiB/s (969MB/s)(423MiB/458msec)

    slat (nsec): min=380, max=97285, avg=2778.70, stdev=1563.26
    clat (usec): min=14, max=23014, avg=63.83, stdev=282.43
     lat (usec): min=17, max=23051, avg=66.61, stdev=282.51
    clat percentiles (usec):
     |  1.00th=[   31],  5.00th=[   35], 10.00th=[   40], 20.00th=[   49],
     | 30.00th=[   53], 40.00th=[   57], 50.00th=[   59], 60.00th=[   62],
     | 70.00th=[   64], 80.00th=[   68], 90.00th=[   75], 95.00th=[   85],
     | 99.00th=[  111], 99.50th=[  122], 99.90th=[  889], 99.95th=[ 1483],
     | 99.99th=[22938]
  write: IOPS=144k, BW=564MiB/s (591MB/s)(1024MiB/1817msec); 0 zone resets
    slat (nsec): min=1152, max=429216, avg=4099.27, stdev=2026.69
    clat (usec): min=11, max=19407, avg=106.50, stdev=458.31
     lat (usec): min=14, max=19411, avg=110.60, stdev=458.36
    clat percentiles (usec):
     |  1.00th=[   35],  5.00th=[   42], 10.00th=[   47], 20.00th=[   57],
     | 30.00th=[   64], 40.00th=[   69], 50.00th=[   73], 60.00th=[   77],
     | 70.00th=[   82], 80.00th=[   91], 90.00th=[  113], 95.00th=[  137],
     | 99.00th=[  685], 99.50th=[ 1500], 99.90th=[ 6915], 99.95th=[12649],
     | 99.99th=[16581]
   bw (  KiB/s): min=379328, max=626696, per=90.85%, avg=524288.00, stdev=105971.00, samples=4
   iops        : min=94832, max=156674, avg=131072.00, stdev=26492.75, samples=4
  lat (usec)   : 20=0.02%, 50=15.72%, 100=73.22%, 250=9.76%, 500=0.39%
  lat (usec)   : 750=0.21%, 1000=0.15%
  lat (msec)   : 2=0.27%, 4=0.14%, 10=0.08%, 20=0.05%, 50=0.01%
  cpu          : usr=33.29%, sys=38.70%, ctx=21042, majf=0, minf=4559
  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=108396,262144,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=16

Run status group 0 (all jobs):
READ: bw=925MiB/s (969MB/s), 925MiB/s-925MiB/s (969MB/s-969MB/s), io=423MiB (444MB), run=458-458msec
WRITE: bw=564MiB/s (591MB/s), 564MiB/s-564MiB/s (591MB/s-591MB/s), io=1024MiB (1074MB), run=1817-1817msec

Disk stats (read/write):

    bcache0: ios=78845/262144, merge=0/0, ticks=1327/10762, in_queue=12089, util=95.41%, aggrios=54450/263725, aggrmerge=0/0, aggrticks=376/2102, aggrin_queue=2479, aggrutil=90.70%

nullb1: ios=105255/265306, merge=0/0, ticks=739/879, in_queue=1619, util=90.70%
nullb0: ios=3646/262144, merge=0/0, ticks=13/3326, in_queue=3340, util=83.61%
+ fio fio/randwrite.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0
RANDWRITE: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=2
...
fio-3.34
Starting 48 processes
Jobs: 48 (f=48): [w(48)][100.0%][w=764MiB/s][w=196k IOPS][eta 00m:00s]
RANDWRITE: (groupid=0, jobs=48): err= 0: pid=9341: Fri May 12 02:36:18 2023
write: IOPS=196k, BW=766MiB/s (803MB/s)(44.9GiB/60003msec); 0 zone resets

    slat (nsec): min=421, max=4604.7k, avg=3817.54, stdev=7945.41
    clat (nsec): min=271, max=82444k, avg=484832.20, stdev=1610621.78
     lat (usec): min=10, max=82448, avg=488.65, stdev=1610.67
    clat percentiles (usec):
     |  1.00th=[   35],  5.00th=[   65], 10.00th=[  118], 20.00th=[  153],
     | 30.00th=[  192], 40.00th=[  233], 50.00th=[  265], 60.00th=[  302],
     | 70.00th=[  359], 80.00th=[  445], 90.00th=[  652], 95.00th=[ 1172],
     | 99.00th=[ 2868], 99.50th=[18220], 99.90th=[22152], 99.95th=[23200],
     | 99.99th=[25560]
   bw (  KiB/s): min=619732, max=1020653, per=100.00%, avg=785206.50, stdev=1535.24, samples=5712
   iops        : min=154933, max=255162, avg=196300.83, stdev=383.81, samples=5712
  lat (nsec)   : 500=0.01%, 750=0.01%, 1000=0.01%
  lat (usec)   : 2=0.01%, 4=0.01%, 10=0.01%, 20=0.07%, 50=3.23%
  lat (usec)   : 100=4.59%, 250=37.78%, 500=38.49%, 750=7.77%, 1000=2.29%
  lat (msec)   : 2=3.49%, 4=1.60%, 10=0.10%, 20=0.30%, 50=0.27%
  lat (msec)   : 100=0.01%
  cpu          : usr=3.09%, sys=7.08%, ctx=10143891, majf=0, minf=604
  IO depths    : 1=0.1%, 2=100.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=0,11763871,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=2

Run status group 0 (all jobs):
WRITE: bw=766MiB/s (803MB/s), 766MiB/s-766MiB/s (803MB/s-803MB/s), io=44.9GiB (48.2GB), run=60003-60003msec

Disk stats (read/write):

    bcache0: ios=88/11739287, merge=0/0, ticks=1/4439670, in_queue=4439671, util=99.53%, aggrios=478/11792831, aggrmerge=0/0, aggrticks=12/507616, aggrin_queue=507628, aggrutil=96.79%

nullb1: ios=821/11821791, merge=0/0, ticks=21/71099, in_queue=71119, util=96.79%
nullb0: ios=135/11763871, merge=0/0, ticks=4/944133, in_queue=944138, util=77.31%
+ for i in 1 2 3
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-default-nowait-off-1.fio
+ for i in 1 2 3 [r(48)][100.0%][r=1963MiB/s][r=503k IOPS][eta 00m:00s]
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-default-nowait-off-2.fio
+ for i in 1 2 3 [r(48)][100.0%][r=1988MiB/s][r=509k IOPS][eta 00m:00s]
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-default-nowait-off-3.fio
+ bcache unregister /dev/nullb0][r=1896MiB/s][r=485k IOPS][eta 00m:00s]
+ bcache unregister /dev/nullb1
+ rmmod bcache.ko
+ rmmod null_blk
+ modprobe null_blk queue_mode=2 nr_devices=2 memory_backed=1 gb=1
+ insmod drivers/md/bcache/bcache.ko bch_nowait=1
+ test_bcache nowait-on
+ bcache make -B /dev/nullb0 -C /dev/nullb1
Name /dev/nullb1
Label
Type cache
UUID: 3960cc68-78c3-4e42-8a72-29a19ec6a97b
Set UUID: bf11b3de-6bc8-42a5-8fd4-66ad7304b9b3
version: 0
nbuckets: 2048
block_size_in_sectors: 1
bucket_size_in_sectors: 1024
nr_in_set: 1
nr_this_dev: 0
first_bucket: 1

                                ...

Name /dev/nullb0
Label
Type data
UUID: 451f8df6-b94c-478c-bd2f-f959f32593c8
Set UUID: bf11b3de-6bc8-42a5-8fd4-66ad7304b9b3
version: 1
block_size_in_sectors: 1
data_offset_in_sectors: 16

+ sleep 1
+ echo /dev/nullb0
+ echo /dev/nullb1
+ dmesg -c

[ 2679.514049] bcache: bcache_device_free() bcache0 stopped
[ 2679.515264] bcache: cache_set_free() Cache set 614c08a8-975f-4eb4-a047-b589e586ffeb unregistered
[ 2679.596435] null_blk: disk nullb0 created
[ 2679.597466] null_blk: disk nullb1 created
[ 2679.597468] null_blk: module loaded
[ 2680.630189] bcache: register_bdev() registered backing device nullb0
[ 2680.631427] bcache: run_cache_set() invalidating existing data
[ 2680.634930] bcache: bch_cached_dev_run() cached dev nullb0 is running already
[ 2680.634945] bcache: bch_cached_dev_attach() Caching nullb0 as bcache0 on set bf11b3de-6bc8-42a5-8fd4-66ad7304b9b3
[ 2680.634959] bcache: register_cache() registered cache device nullb1
+ lsblk
NAME      MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
sda         8:0    0   50G  0 disk 
├─sda1      8:1    0    1G  0 part /boot
└─sda2      8:2    0   49G  0 part /home
sdb         8:16   0  100G  0 disk /mnt/data
sr0        11:0    1 1024M  0 rom  
nullb0    251:0    0    1G  0 disk 
└─bcache0 250:0    0 1024M  0 disk 
nullb1    251:1    0    1G  0 disk 
└─bcache0 250:0    0 1024M  0 disk 
vda       252:0    0    5G  0 disk 
nvme0n1   259:0    0    1G  0 disk 

+ sleep 1
+ fio fio/verify.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0
write-and-verify: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=16 fio-3.34
Starting 1 process
Jobs: 1 (f=1): [V(1)][57.1%][r=5089KiB/s,w=251MiB/s][r=1272,w=64.4k IOPS][eta 00m:03s]
write-and-verify: (groupid=0, jobs=1): err= 0: pid=9962: Fri May 12 02:39:25 2023
read: IOPS=203k, BW=792MiB/s (830MB/s)(647MiB/817msec)

    slat (usec): min=2, max=4887, avg= 4.01, stdev=20.39
    clat (nsec): min=80, max=5611.0k, avg=74109.33, stdev=92291.20
     lat (usec): min=2, max=5623, avg=78.12, stdev=95.01
    clat percentiles (usec):
     |  1.00th=[   16],  5.00th=[   59], 10.00th=[   60], 20.00th=[   61],
     | 30.00th=[   63], 40.00th=[   67], 50.00th=[   68], 60.00th=[   70],
     | 70.00th=[   73], 80.00th=[   76], 90.00th=[   83], 95.00th=[   98],
     | 99.00th=[  143], 99.50th=[  174], 99.90th=[ 1713], 99.95th=[ 2040],
     | 99.99th=[ 2278]
  write: IOPS=67.4k, BW=263MiB/s (276MB/s)(1024MiB/3887msec); 0 zone resets
    slat (usec): min=4, max=4882, avg=12.71, stdev=39.18
    clat (nsec): min=161, max=106211k, avg=224208.69, stdev=951424.56
     lat (usec): min=9, max=106224, avg=236.92, stdev=952.58
    clat percentiles (usec):
     |  1.00th=[   87],  5.00th=[  128], 10.00th=[  137], 20.00th=[  143],
     | 30.00th=[  149], 40.00th=[  159], 50.00th=[  176], 60.00th=[  184],
     | 70.00th=[  192], 80.00th=[  210], 90.00th=[  269], 95.00th=[  322],
     | 99.00th=[ 1029], 99.50th=[ 1614], 99.90th=[ 9110], 99.95th=[14091],
     | 99.99th=[21627]
   bw (  KiB/s): min=185096, max=314112, per=97.18%, avg=262144.00, stdev=45100.35, samples=8
   iops        : min=46274, max=78528, avg=65536.00, stdev=11275.09, samples=8
  lat (nsec)   : 100=0.03%, 250=0.25%, 500=0.01%, 750=0.01%, 1000=0.01%
  lat (usec)   : 4=0.02%, 10=0.05%, 20=0.08%, 50=0.43%, 100=36.75%
  lat (usec)   : 250=54.54%, 500=6.33%, 750=0.41%, 1000=0.38%
  lat (msec)   : 2=0.48%, 4=0.12%, 10=0.06%, 20=0.04%, 50=0.01%
  lat (msec)   : 250=0.01%
  cpu          : usr=17.58%, sys=54.71%, ctx=260303, majf=0, minf=4542
  IO depths    : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=165554,262144,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=16

Run status group 0 (all jobs):
READ: bw=792MiB/s (830MB/s), 792MiB/s-792MiB/s (830MB/s-830MB/s), io=647MiB (678MB), run=817-817msec
WRITE: bw=263MiB/s (276MB/s), 263MiB/s-263MiB/s (276MB/s-276MB/s), io=1024MiB (1074MB), run=3887-3887msec

Disk stats (read/write):

    bcache0: ios=152330/262144, merge=0/0, ticks=618/10341, in_queue=10959, util=97.68%, aggrios=83009/263342, aggrmerge=0/0, aggrticks=128/1245, aggrin_queue=1373, aggrutil=90.86%

nullb1: ios=155789/264540, merge=0/0, ticks=229/623, in_queue=852, util=90.86%
nullb0: ios=10229/262144, merge=0/0, ticks=28/1867, in_queue=1895, util=89.00%
+ fio fio/randwrite.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0
RANDWRITE: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=2
...
fio-3.34
Starting 48 processes
Jobs: 46 (f=46): [w(35),_(1),w(4),_(1),w(7)][95.3%][w=762MiB/s][w=195k IOPS][eta 00m:03s]
RANDWRITE: (groupid=0, jobs=48): err= 0: pid=10046: Fri May 12 02:40:26 2023
write: IOPS=206k, BW=804MiB/s (843MB/s)(47.1GiB/60016msec); 0 zone resets

    slat (nsec): min=1002, max=24597k, avg=146865.06, stdev=202132.32
    clat (nsec): min=110, max=83997k, avg=318946.95, stdev=1533223.26
     lat (usec): min=7, max=84012, avg=465.81, stdev=1551.87
    clat percentiles (nsec):
     |  1.00th=[     322],  5.00th=[   16512], 10.00th=[   21120],
     | 20.00th=[   65280], 30.00th=[  100864], 40.00th=[  110080],
     | 50.00th=[  116224], 60.00th=[  123392], 70.00th=[  136192],
     | 80.00th=[  162816], 90.00th=[  329728], 95.00th=[  790528],
     | 99.00th=[ 2867200], 99.50th=[17432576], 99.90th=[21364736],
     | 99.95th=[22413312], 99.99th=[24510464]
   bw (  KiB/s): min=483303, max=1030912, per=100.00%, avg=824214.25, stdev=2303.88, samples=5710
   iops        : min=120824, max=257728, avg=206051.21, stdev=575.97, samples=5710
  lat (nsec)   : 250=0.54%, 500=1.46%, 750=0.45%, 1000=0.14%
  lat (usec)   : 2=0.05%, 4=0.01%, 10=0.01%, 20=6.20%, 50=9.97%
  lat (usec)   : 100=10.80%, 250=58.00%, 500=4.66%, 750=2.61%, 1000=0.49%
  lat (msec)   : 2=2.36%, 4=1.60%, 10=0.08%, 20=0.38%, 50=0.19%
  lat (msec)   : 100=0.01%
  cpu          : usr=0.53%, sys=51.34%, ctx=12806805, majf=0, minf=671
  IO depths    : 1=0.1%, 2=100.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=0,12345076,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=2

Run status group 0 (all jobs):
WRITE: bw=804MiB/s (843MB/s), 804MiB/s-804MiB/s (843MB/s-843MB/s), io=47.1GiB (50.6GB), run=60016-60016msec

Disk stats (read/write):

    bcache0: ios=176/12313695, merge=0/0, ticks=13/3916805, in_queue=3916818, util=98.25%, aggrios=959/12338321, aggrmerge=0/0, aggrticks=55/222368, aggrin_queue=222424, aggrutil=96.02%

nullb1: ios=1641/12331567, merge=0/0, ticks=86/51128, in_queue=51214, util=96.02%
nullb0: ios=277/12345076, merge=0/0, ticks=24/393609, in_queue=393634, util=76.64%
+ for i in 1 2 3
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-nowait-on-1.fio
+ for i in 1 2 3 [r(48)][100.0%][r=2086MiB/s][r=534k IOPS][eta 00m:00s]
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-nowait-on-2.fio
+ for i in 1 2 3 [r(48)][100.0%][r=2185MiB/s][r=559k IOPS][eta 00m:00s]
+ fio fio/randread.fio --ioengine=io_uring --size=1G --filename=/dev/bcache0 --output=bc-nowait-on-3.fio
+ bcache unregister /dev/nullb0][r=2148MiB/s][r=550k IOPS][eta 00m:00s]
+ bcache unregister /dev/nullb1
+ rmmod bcache.ko
+ rmmod null_blk
linux-block (for-next) # for i in IOPS slat cpu; do grep $i bc-*fio | column -t ; done

bc-default-nowait-off-1.fio:  read:  IOPS=482k,  BW=1885MiB/s  (1976MB/s)(110GiB/60003msec)
bc-default-nowait-off-2.fio:  read:  IOPS=484k,  BW=1889MiB/s  (1981MB/s)(111GiB/60002msec)
bc-default-nowait-off-3.fio:  read:  IOPS=483k,  BW=1886MiB/s  (1978MB/s)(111GiB/60002msec)
bc-nowait-on-1.fio:           read:  IOPS=544k,  BW=2125MiB/s  (2228MB/s)(124GiB/60001msec)
bc-nowait-on-2.fio:           read:  IOPS=547k,  BW=2137MiB/s  (2241MB/s)(125GiB/60002msec)
bc-nowait-on-3.fio:           read:  IOPS=546k,  BW=2132MiB/s  (2236MB/s)(125GiB/60001msec)
bc-default-nowait-off-1.fio:  slat  (nsec):  min=430,  max=5488.5k,  avg=2797.52,  stdev=5775.84
bc-default-nowait-off-2.fio:  slat  (nsec):  min=431,  max=8252.4k,  avg=2805.33,  stdev=6216.69
bc-default-nowait-off-3.fio:  slat  (nsec):  min=431,  max=6846.6k,  avg=2814.57,  stdev=6309.09
bc-nowait-on-1.fio:           slat  (usec):  min=2,    max=39086,    avg=87.48,    stdev=427.87
bc-nowait-on-2.fio:           slat  (usec):  min=3,    max=39519,    avg=86.98,    stdev=425.06
bc-nowait-on-3.fio:           slat  (usec):  min=3,    max=38880,    avg=87.17,    stdev=427.17
bc-default-nowait-off-1.fio:  cpu  :  usr=2.77%,  sys=6.57%,   ctx=22015526,  majf=0,  minf=697
bc-default-nowait-off-2.fio:  cpu  :  usr=2.75%,  sys=6.59%,   ctx=22003700,  majf=0,  minf=666
bc-default-nowait-off-3.fio:  cpu  :  usr=2.81%,  sys=6.57%,   ctx=21938309,  majf=0,  minf=659
bc-nowait-on-1.fio:           cpu  :  usr=1.08%,  sys=78.39%,  ctx=2744092,   majf=0,  minf=794
bc-nowait-on-2.fio:           cpu  :  usr=1.10%,  sys=79.76%,  ctx=2537466,   majf=0,  minf=787
bc-nowait-on-3.fio:           cpu  :  usr=1.10%,  sys=79.88%,  ctx=2528092,   majf=0,  minf=788

linux-block (for-next) #

#1 lore raw

PATCH 1/1 bcache: allow user to set QUEUE_FLAG_NOWAIT

Allow user to set the QUEUE_FLAG_NOWAIT optionally using module parameter to retain the default behaviour. Also, update respective allocation flags in the write path. Following are the performance numbers with io_uring fio engine for random read, note that device has been populated fully with randwrite workload before taking these numbers :-

* linux-block (for-next) # grep IOPS bc-*fio | column -t

nowait-off-1.fio:  read:  IOPS=482k,  BW=1885MiB/s
nowait-off-2.fio:  read:  IOPS=484k,  BW=1889MiB/s
nowait-off-3.fio:  read:  IOPS=483k,  BW=1886MiB/s
nowait-on-1.fio:   read:  IOPS=544k,  BW=2125MiB/s
nowait-on-2.fio:   read:  IOPS=547k,  BW=2137MiB/s
nowait-on-3.fio:   read:  IOPS=546k,  BW=2132MiB/s

* linux-block (for-next) # grep slat bc-*fio | column -t

nowait-off-1.fio: slat (nsec): min=430, max=5488.5k, avg=2797.52 nowait-off-2.fio: slat (nsec): min=431, max=8252.4k, avg=2805.33 nowait-off-3.fio: slat (nsec): min=431, max=6846.6k, avg=2814.57

nowait-on-1.fio:  slat (usec):  min=2,   max=39086,   avg=87.48
nowait-on-2.fio:  slat (usec):  min=3,   max=39519,   avg=86.98
nowait-on-3.fio:  slat (usec):  min=3,   max=38880,   avg=87.17

* linux-block (for-next) # grep cpu bc-*fio | column -t

nowait-off-1.fio:  cpu  :  usr=2.77%,  sys=6.57%,   ctx=22015526
nowait-off-2.fio:  cpu  :  usr=2.75%,  sys=6.59%,   ctx=22003700
nowait-off-3.fio:  cpu  :  usr=2.81%,  sys=6.57%,   ctx=21938309
nowait-on-1.fio:   cpu  :  usr=1.08%,  sys=78.39%,  ctx=2744092
nowait-on-2.fio:   cpu  :  usr=1.10%,  sys=79.76%,  ctx=2537466
nowait-on-3.fio:   cpu  :  usr=1.10%,  sys=79.88%,  ctx=2528092
---
 drivers/md/bcache/request.c | 3 ++-
 drivers/md/bcache/super.c   | 6 ++++++
 2 files changed, 8 insertions(+), 1 deletion(-)
drivers/md/bcache/request.c +2 -1
index 67a2e29e0b40..2055a23eb4b7 100644
@@ -716,9 +716,10 @@ static inline struct search *search_alloc(struct bio *bio,
 		struct bcache_device *d, struct block_device *orig_bdev,
 		unsigned long start_time)
 {
+	gfp_t gfp = bio->bi_opf & REQ_NOWAIT ? GFP_NOWAIT : GFP_NOIO;
 	struct search *s;
 
-	s = mempool_alloc(&d->c->search, GFP_NOIO);
+	s = mempool_alloc(&d->c->search, gfp);
 
 	closure_init(&s->cl, NULL);
 	do_bio_hook(s, bio, request_endio);
drivers/md/bcache/super.c +6 -0
index 7e9d19fd21dd..f76822bece26 100644
@@ -28,6 +28,7 @@
 
 unsigned int bch_cutoff_writeback;
 unsigned int bch_cutoff_writeback_sync;
+bool bch_nowait;
 
 static const char bcache_magic[] = {
 	0xc6, 0x85, 0x73, 0xf6, 0x4e, 0x1a, 0x45, 0xca,
@@ -971,6 +972,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
 	}
 
 	blk_queue_flag_set(QUEUE_FLAG_NONROT, d->disk->queue);
+	if (bch_nowait)
+		blk_queue_flag_set(QUEUE_FLAG_NOWAIT, d->disk->queue);
 
 	blk_queue_write_cache(q, true, true);
 
@@ -2933,6 +2936,9 @@ MODULE_PARM_DESC(bch_cutoff_writeback, "threshold to cutoff writeback");
 module_param(bch_cutoff_writeback_sync, uint, 0);
 MODULE_PARM_DESC(bch_cutoff_writeback_sync, "hard threshold to cutoff writeback");
 
+module_param(bch_nowait, bool, 0);
+MODULE_PARM_DESC(bch_nowait, "set QUEUE_FLAG_NOWAIT");
+
 MODULE_DESCRIPTION("Bcache: a Linux block layer cache");
 MODULE_AUTHOR("Kent Overstreet <[email protected]>");
 MODULE_LICENSE("GPL");
Signature · 1 line
2.40.0
#2 lore raw
Quoted 29 lines · Chaitanya Kulkarni

Allow user to set the QUEUE_FLAG_NOWAIT optionally using module parameter to retain the default behaviour. Also, update respective allocation flags in the write path. Following are the performance numbers with io_uring fio engine for random read, note that device has been populated fully with randwrite workload before taking these numbers :-

* linux-block (for-next) # grep IOPS bc-*fio | column -t

nowait-off-1.fio:  read:  IOPS=482k,  BW=1885MiB/s
nowait-off-2.fio:  read:  IOPS=484k,  BW=1889MiB/s
nowait-off-3.fio:  read:  IOPS=483k,  BW=1886MiB/s
nowait-on-1.fio:   read:  IOPS=544k,  BW=2125MiB/s
nowait-on-2.fio:   read:  IOPS=547k,  BW=2137MiB/s
nowait-on-3.fio:   read:  IOPS=546k,  BW=2132MiB/s

* linux-block (for-next) # grep slat bc-*fio | column -t

nowait-off-1.fio: slat (nsec): min=430, max=5488.5k, avg=2797.52 nowait-off-2.fio: slat (nsec): min=431, max=8252.4k, avg=2805.33 nowait-off-3.fio: slat (nsec): min=431, max=6846.6k, avg=2814.57

nowait-on-1.fio:  slat (usec):  min=2,   max=39086,   avg=87.48
nowait-on-2.fio:  slat (usec):  min=3,   max=39519,   avg=86.98
nowait-on-3.fio:  slat (usec):  min=3,   max=38880,   avg=87.17

* linux-block (for-next) # grep cpu bc-*fio | column -t

Quoted 7 lines
nowait-off-1.fio:  cpu  :  usr=2.77%,  sys=6.57%,   ctx=22015526
nowait-off-2.fio:  cpu  :  usr=2.75%,  sys=6.59%,   ctx=22003700
nowait-off-3.fio:  cpu  :  usr=2.81%,  sys=6.57%,   ctx=21938309
nowait-on-1.fio:   cpu  :  usr=1.08%,  sys=78.39%,  ctx=2744092
nowait-on-2.fio:   cpu  :  usr=1.10%,  sys=79.76%,  ctx=2537466
nowait-on-3.fio:   cpu  :  usr=1.10%,  sys=79.88%,  ctx=2528092

Wow, amazing for such a tiny patch. Especially the latency numbers! Given this, maybe NOWAIT should be enabled by default.

Why would anyone want to use the old NOWAIT=off variant?

Are there benefits to going without NOWAIT that go unnoticed when testing against a ramdisk?

For example, (and this seems unlikely) can NOWAIT affect the IO scheduler in a way that would prevent sorted IOs headed toward a rotational disk?

It would be interesting to see two more test classes:

1. Ram disk cache with NVMe backing device.

2. NVMe cache with rotational HDD backing device.

-Eric

Quoted 58 lines
---
 drivers/md/bcache/request.c | 3 ++-
 drivers/md/bcache/super.c   | 6 ++++++
 2 files changed, 8 insertions(+), 1 deletion(-)
drivers/md/bcache/request.c +2 -1
index 67a2e29e0b40..2055a23eb4b7 100644
@@ -716,9 +716,10 @@ static inline struct search *search_alloc(struct bio *bio,
 		struct bcache_device *d, struct block_device *orig_bdev,
 		unsigned long start_time)
 {
+	gfp_t gfp = bio->bi_opf & REQ_NOWAIT ? GFP_NOWAIT : GFP_NOIO;
 	struct search *s;
 
-	s = mempool_alloc(&d->c->search, GFP_NOIO);
+	s = mempool_alloc(&d->c->search, gfp);
 
 	closure_init(&s->cl, NULL);
 	do_bio_hook(s, bio, request_endio);
drivers/md/bcache/super.c +6 -0
index 7e9d19fd21dd..f76822bece26 100644
@@ -28,6 +28,7 @@
 
 unsigned int bch_cutoff_writeback;
 unsigned int bch_cutoff_writeback_sync;
+bool bch_nowait;
 
 static const char bcache_magic[] = {
 	0xc6, 0x85, 0x73, 0xf6, 0x4e, 0x1a, 0x45, 0xca,
@@ -971,6 +972,8 @@ static int bcache_device_init(struct bcache_device *d, unsigned int block_size,
 	}
 
 	blk_queue_flag_set(QUEUE_FLAG_NONROT, d->disk->queue);
+	if (bch_nowait)
+		blk_queue_flag_set(QUEUE_FLAG_NOWAIT, d->disk->queue);
 
 	blk_queue_write_cache(q, true, true);
 
@@ -2933,6 +2936,9 @@ MODULE_PARM_DESC(bch_cutoff_writeback, "threshold to cutoff writeback");
 module_param(bch_cutoff_writeback_sync, uint, 0);
 MODULE_PARM_DESC(bch_cutoff_writeback_sync, "hard threshold to cutoff writeback");
 
+module_param(bch_nowait, bool, 0);
+MODULE_PARM_DESC(bch_nowait, "set QUEUE_FLAG_NOWAIT");
+
 MODULE_DESCRIPTION("Bcache: a Linux block layer cache");
 MODULE_AUTHOR("Kent Overstreet <[email protected]>");
 MODULE_LICENSE("GPL");
Signature · 1 line
2.40.0
#3 lore raw
Quoted 35 lines · Eric Wheeler
Quoted 33 lines · Chaitanya Kulkarni

Allow user to set the QUEUE_FLAG_NOWAIT optionally using module parameter to retain the default behaviour. Also, update respective allocation flags in the write path. Following are the performance numbers with io_uring fio engine for random read, note that device has been populated fully with randwrite workload before taking these numbers :-

* linux-block (for-next) # grep IOPS bc-*fio | column -t

nowait-off-1.fio:  read:  IOPS=482k,  BW=1885MiB/s
nowait-off-2.fio:  read:  IOPS=484k,  BW=1889MiB/s
nowait-off-3.fio:  read:  IOPS=483k,  BW=1886MiB/s
nowait-on-1.fio:   read:  IOPS=544k,  BW=2125MiB/s
nowait-on-2.fio:   read:  IOPS=547k,  BW=2137MiB/s
nowait-on-3.fio:   read:  IOPS=546k,  BW=2132MiB/s

* linux-block (for-next) # grep slat bc-*fio | column -t

nowait-off-1.fio: slat (nsec): min=430, max=5488.5k, avg=2797.52 nowait-off-2.fio: slat (nsec): min=431, max=8252.4k, avg=2805.33 nowait-off-3.fio: slat (nsec): min=431, max=6846.6k, avg=2814.57

nowait-on-1.fio:  slat (usec):  min=2,   max=39086,   avg=87.48
nowait-on-2.fio:  slat (usec):  min=3,   max=39519,   avg=86.98
nowait-on-3.fio:  slat (usec):  min=3,   max=38880,   avg=87.17

* linux-block (for-next) # grep cpu bc-*fio | column -t

nowait-off-1.fio:  cpu  :  usr=2.77%,  sys=6.57%,   ctx=22015526
nowait-off-2.fio:  cpu  :  usr=2.75%,  sys=6.59%,   ctx=22003700
nowait-off-3.fio:  cpu  :  usr=2.81%,  sys=6.57%,   ctx=21938309
Quoted 6 lines
Quoted 3 lines
nowait-on-1.fio:   cpu  :  usr=1.08%,  sys=78.39%,  ctx=2744092
nowait-on-2.fio:   cpu  :  usr=1.10%,  sys=79.76%,  ctx=2537466
nowait-on-3.fio:   cpu  :  usr=1.10%,  sys=79.88%,  ctx=2528092

Wow, amazing for such a tiny patch. Especially the latency numbers! Given this, maybe NOWAIT should be enabled by default.

yes, sending out V2 without the module parameter as my other patches has similar comment to enable nowait by default.

Quoted 1 line

Why would anyone want to use the old NOWAIT=off variant?

for some reason I wanted to make it backward compatible just in case this change creates regression on the setup that I don't have access to ..

Quoted 2 lines

Are there benefits to going without NOWAIT that go unnoticed when testing against a ramdisk?

no not at all from what I can see ramdisk enables it by default ..

Quoted 2 lines

For example, (and this seems unlikely) can NOWAIT affect the IO scheduler in a way that would prevent sorted IOs headed toward a rotational disk?

unfortunately I don't have a setup to run those tests :(.

Quoted 9 lines

It would be interesting to see two more test classes:

1. Ram disk cache with NVMe backing device.

2. NVMe cache with rotational HDD backing device.

-Eric

once I get a setup I'll be happy to run those numbers ..

-ck