Hi Chaitanya
Sorry for the late response, as it take me some time to bisect this
issue, seems it was introduced with the below commit:
23f3e3272e7a4d9fb870485cd6df1e4f9539282c
Author: Xiao Ni <xni@xxxxxxxxxx>
Date: Thu Feb 9 11:19:30 2023 +0800
block: Merge bio before checking ->cached_rq
Here are the steps I used to reproduce it:
1. do fio stress test on three nvme disks background
2. offline the first three cpus
# echo 0 > /sys/devices/system/cpu/cpu1/online
# echo 0 > /sys/devices/system/cpu/cpu2/online
# echo 0 > /sys/devices/system/cpu/cpu3/online
3. update each disk's scheduler/nr_requests
for disk in nvme0n1 nvme1n1 nvme2n1; do
{
scheds="mq-deadline kyber bfq"
for sched in $scheds; do
echo $sched > /sys/block/$disk/queue/scheduler
max_nr="$(cat /sys/block/$disk/queue/nr_requests)"
for ((nr = 4; nr <= max_nr; nr++)); do
echo $nr > /sys/block/$disk/queue/nr_requests
done
done
} &
done
4. online the cpus
echo 1 > /sys/devices/system/cpu/cpu1/online
echo 1 > /sys/devices/system/cpu/cpu2/online
echo 1 > /sys/devices/system/cpu/cpu3/online
On Sun, Apr 16, 2023 at 12:52 PM Chaitanya Kulkarni
<chaitanyak@xxxxxxxxxx> wrote:
>
> On 4/14/23 08:04, Yi Zhang wrote:
> > Hello
> >
> > Bellow panic triggered during nvme disks queue/nr_request update with
> > fio background, pls help check it and let me know if you need any
> > info/test, thanks.
> >
> >
>
>
> Is it by any specific blktests ?
>
> can you please share the steps ?
>
> -ck
>
>
--
Best Regards,
Yi Zhang
