Re: [PATCH blktests] nvmeof-mp/001: fix failure when CONFIG_NVME_HWMON enabled

[Omar Sandoval <osandov@osandov.com>]

On Sun, Sep 12, 2021 at 08:03:36PM +0800, Yi Zhang wrote:
> Skip checking ng0n1/hwmon5 in count_devices
> 
> $ use_siw=1  ./check nvmeof-mp/001
> nvmeof-mp/001 (Log in and log out)                           [failed]
>     runtime  3.695s  ...  4.002s
>     --- tests/nvmeof-mp/001.out	2021-09-12 05:35:17.866892187 -0400
>     +++ /root/blktests/results/nodev/nvmeof-mp/001.out.bad	2021-09-12 06:49:25.621880616 -0400
>     @@ -1,3 +1,3 @@
>      Configured NVMe target driver
>     -count_devices(): 1 <> 1
>     +count_devices(): 3 <> 1
>      Passed
> $ ls -l /sys/class/nvme-fabrics/ctl/*/*/device
> lrwxrwxrwx. 1 root root 0 Sep 12 06:49 /sys/class/nvme-fabrics/ctl/nvme0/hwmon5/device -> ../../nvme0
> lrwxrwxrwx. 1 root root 0 Sep 12 06:49 /sys/class/nvme-fabrics/ctl/nvme0/ng0n1/device -> ../../nvme0
> lrwxrwxrwx. 1 root root 0 Sep 12 06:49 /sys/class/nvme-fabrics/ctl/nvme0/nvme0n1/device -> ../../nvme0
> 
> Signed-off-by: Yi Zhang <yi.zhang@redhat.com>
> ---
>  tests/nvmeof-mp/001 | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/tests/nvmeof-mp/001 b/tests/nvmeof-mp/001
> index 69f1e24..3c3e8d5 100755
> --- a/tests/nvmeof-mp/001
> +++ b/tests/nvmeof-mp/001
> @@ -11,6 +11,7 @@ count_devices() {
>  	local d devs=0
>  
>  	for d in /sys/class/nvme-fabrics/ctl/*/*/device; do
> +		[[ "$d" =~ hwmon[0-9]|ng[0-9]n[0-9] ]] && continue

I'm not too familiar with NVME_HWMON, but presumably it's possible to
have more than 10 of these "ng" devices. Should this be ng[0-9]+n[0-9]?

>  		[ -d "$d" ] && ((devs++))
>  	done
>  	echo "$devs"
> -- 
> 2.21.3
>