[linus:master] [btrfs] 4e159150a9: fio.write_iops 6.5% regression

[kernel test robot]

Hello,

kernel test robot noticed a 6.5% regression of fio.write_iops on:


commit: 4e159150a9a56d66d247f4b5510bed46fe58aa1c ("btrfs: do not strictly require dirty metadata threshold for metadata writepages")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master

[still regression on linus/master      6589b3d76db2d6adbf8f2084c303fb24252a0dc6]
[still regression on linux-next/master fd9678829d6dd0c10fde080b536abf4b1121c346]

testcase: fio-basic
config: x86_64-rhel-9.4
compiler: gcc-14
test machine: 64 threads 2 sockets Intel(R) Xeon(R) Gold 6346 CPU @ 3.10GHz (Ice Lake) with 256G memory
parameters:

	runtime: 300s
	disk: 1HDD
	fs: btrfs
	nr_task: 1
	test_size: 128G
	rw: write
	bs: 4k
	ioengine: pvsync2
	cpufreq_governor: performance


If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
| Closes: https://lore.kernel.org/oe-lkp/202602122319.d71b793c-lkp@xxxxxxxxx


Details are as below:
-------------------------------------------------------------------------------------------------->


The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20260212/202602122319.d71b793c-lkp@xxxxxxxxx

=========================================================================================
bs/compiler/cpufreq_governor/disk/fs/ioengine/kconfig/nr_task/rootfs/runtime/rw/tbox_group/test_size/testcase:
  4k/gcc-14/performance/1HDD/btrfs/pvsync2/x86_64-rhel-9.4/1/debian-13-x86_64-20250902.cgz/300s/write/lkp-icl-2sp9/128G/fio-basic

commit: 
  3430818739 ("btrfs: add extra device item checks at mount")
  4e159150a9 ("btrfs: do not strictly require dirty metadata threshold for metadata writepages")

34308187395ff01f 4e159150a9a56d66d247f4b5510 
---------------- --------------------------- 
         %stddev     %change         %stddev
             \          |                \  
  25998829            -6.1%   24422371        cpuidle..usage
      1.06            +3.2%       1.09        iostat.cpu.iowait
      1215 ±  3%     -10.1%       1093 ±  3%  meminfo.Dirty
     80085            -6.5%      74897        vmstat.io.bo
    121741            -6.4%     113948        vmstat.system.cs
     87339            -6.0%      82083        vmstat.system.in
     15.13 ± 50%     +75.9%      26.62 ± 22%  sched_debug.cfs_rq:/.removed.load_avg.avg
    287273            -5.8%     270535        sched_debug.cpu.nr_switches.avg
   1713834 ± 11%     -19.8%    1374291 ± 11%  sched_debug.cpu.nr_switches.max
    418676 ±  7%     -18.5%     341334 ± 11%  sched_debug.cpu.nr_switches.stddev
     47.67            -5.6%      45.00        turbostat.Avg_MHz
      1.32            -0.1        1.26        turbostat.Busy%
  32589252            -6.1%   30597064        turbostat.IRQ
     94552            -5.0%      89817        turbostat.NMI
   6080930            -6.5%    5686530        proc-vmstat.nr_dirtied
    121315            -3.5%     117069        proc-vmstat.nr_slab_unreclaimable
   6080407            -6.5%    5686054        proc-vmstat.nr_written
   7057391            -5.6%    6658839        proc-vmstat.numa_hit
   6991110            -5.7%    6592460        proc-vmstat.numa_local
   7583554            -5.0%    7205949        proc-vmstat.pgalloc_normal
   7342680            -5.0%    6976953        proc-vmstat.pgfree
  24345737            -6.5%   22768702        proc-vmstat.pgpgout
      0.01            +0.0        0.02 ± 12%  fio.latency_1000us%
      0.01 ± 19%      +0.1        0.09 ± 22%  fio.latency_2ms%
      0.02 ± 23%      +0.1        0.10 ± 21%  fio.latency_4ms%
  48595970            -6.5%   45440610        fio.time.file_system_outputs
     25.67            -7.1%      23.83        fio.time.percent_of_cpu_this_job_got
     71.70            -7.2%      66.52        fio.time.system_time
   6108254            -6.5%    5713199        fio.time.voluntary_context_switches
   6074496            -6.5%    5680076        fio.workload
     79.09            -6.5%      73.96        fio.write_bw_MBps
     48958            +7.0%      52394        fio.write_clat_mean_ns
    250398 ±  2%     +31.1%     328384        fio.write_clat_stddev
     20248            -6.5%      18933        fio.write_iops
 5.513e+08            -4.8%  5.247e+08        perf-stat.i.branch-instructions
  21121264            -5.5%   19957666 ±  2%  perf-stat.i.cache-references
    122833            -6.4%     114967        perf-stat.i.context-switches
 3.438e+09            -4.9%  3.269e+09        perf-stat.i.cpu-cycles
 2.651e+09            -4.8%  2.523e+09        perf-stat.i.instructions
      1.92            -6.4%       1.80        perf-stat.i.metric.K/sec
      1.63            +0.1        1.71        perf-stat.overall.branch-miss-rate%
    131286            +1.8%     133654        perf-stat.overall.path-length
 5.496e+08            -4.8%  5.231e+08        perf-stat.ps.branch-instructions
  21055046            -5.5%   19895494 ±  2%  perf-stat.ps.cache-references
    122422            -6.4%     114585        perf-stat.ps.context-switches
 3.429e+09            -4.9%  3.261e+09        perf-stat.ps.cpu-cycles
 2.643e+09            -4.8%  2.516e+09        perf-stat.ps.instructions
 7.975e+11            -4.8%  7.592e+11        perf-stat.total.instructions
      0.61 ±  6%      +0.1        0.70 ±  5%  perf-profile.calltrace.cycles-pp.try_to_block_task.__schedule.schedule.io_schedule.folio_wait_bit_common
      0.68 ±  6%      +0.1        0.78 ±  8%  perf-profile.calltrace.cycles-pp.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.pv_native_safe_halt.acpi_safe_halt
      0.65 ±  6%      +0.1        0.75 ±  8%  perf-profile.calltrace.cycles-pp.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.pv_native_safe_halt
      0.49 ± 45%      +0.2        0.66 ±  5%  perf-profile.calltrace.cycles-pp.dequeue_task_fair.try_to_block_task.__schedule.schedule.io_schedule
      1.29 ±  5%      +0.2        1.49 ±  8%  perf-profile.calltrace.cycles-pp.io_schedule.folio_wait_bit_common.folio_wait_writeback.__filemap_fdatawait_range.filemap_fdatawait_range
      1.24 ±  6%      +0.2        1.44 ±  8%  perf-profile.calltrace.cycles-pp.schedule.io_schedule.folio_wait_bit_common.folio_wait_writeback.__filemap_fdatawait_range
      0.46 ± 45%      +0.3        0.80 ± 29%  perf-profile.calltrace.cycles-pp.dequeue_entities.dequeue_task_fair.try_to_block_task.__schedule.schedule
      0.20 ±  6%      -0.0        0.16 ± 12%  perf-profile.children.cycles-pp.fio_gettime
      0.08 ± 16%      +0.0        0.11 ± 11%  perf-profile.children.cycles-pp.rmqueue_pcplist
      0.32 ±  6%      +0.0        0.36 ± 10%  perf-profile.children.cycles-pp.native_apic_msr_eoi
      0.74 ±  5%      +0.1        0.83 ±  6%  perf-profile.children.cycles-pp.hrtimer_interrupt
      1.29 ±  5%      +0.2        1.49 ±  8%  perf-profile.children.cycles-pp.io_schedule
      0.20 ±  6%      -0.0        0.16 ± 14%  perf-profile.self.cycles-pp.fio_gettime
      0.12 ± 24%      -0.0        0.09 ± 10%  perf-profile.self.cycles-pp.check_delayed_ref
      0.02 ±141%      +0.0        0.06 ±  9%  perf-profile.self.cycles-pp.btrfs_csum_file_blocks




Disclaimer:
Results have been estimated based on internal Intel analysis and are provided
for informational purposes only. Any difference in system hardware or software
design or configuration may affect actual performance.


-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki