[VROC]Call trace during IO on IMSM RAID 5 (fio performance check)

Bug #1801282 reported by quanxian
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
intel
Won't Fix
Undecided
Unassigned
Ubuntu
New
Undecided
Unassigned

Bug Description

Description

There is a call trace during fio testing of IMSM RAID 5 array. Call trace is attached.

How reproducible: always (after few hours of fio testing)

Steps to Reproduce:
1. Create IMSM RAID5 with 4 disks:
mdadm -C /dev/md/imsm0 -n 4 -e imsm /dev/nvme0n1 /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1
mdadm -C /dev/md126 /dev/md/imsm0 -n4 -l 5 -c 64 -f
2. Increase number of RAID5 threads:
echo 4 > /sys/block/md126/md/group_thread_cnt
3. Trigger fio:
fio --time_based --name="fio_check=16=16=4kb=100" --filename=/dev/md126 --rw=randrw --rwmixwrite=100 --bs=4kb --direct=1 --do_verify=0 --eta=never --iodepth=16 --ioengine=libaio --runtime=600 --ramp_time=600 --refill_buffers --numjobs=16 --group_reporting --randrepeat=0 --thread --unified_rw_reporting=1 --cpus_allowed=2-17 --cpus_allowed_policy=split --disable_lat=1 --disable_slat=1 --disable_bw=1 --stonewall --norandommap --minimal --output=results_16_16_4kb_100.log

Target Kernel: TBD
Target Release: 19.10

quanxian (quanxian-wang)
tags: added: intel-kernel-19.10
removed: intel-kernel-19.04
description: updated
quanxian (quanxian-wang)
Changed in intel:
status: New → Won't Fix
Revision history for this message
quanxian (quanxian-wang) wrote :

if reproduce, just reopen it.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.