Re: [PATCH blktests] common/multipath-over-rdma: fix warning ignored null byte in input

[Bart Van Assche <bvanassche@acm.org>]

On 2020-08-19 08:16, Yi Zhang wrote:
> [blktests]# nvme_trtype=rdma ./check nvme/004
> nvme/004 (test nvme and nvmet UUID NS descriptors)
>     runtime  1.238s  ...
> nvme/004 (test nvme and nvmet UUID NS descriptors)           [failed]ignored null byte in input
>     runtime  1.238s  ...  1.237s 410: warning: command substitution: ignored null byte in input
>     --- tests/nvme/004.out	2020-08-18 08:11:08.496809985 -0400
>     +++ /root/blktests/results/nodev/nvme/004.out.bad	2020-08-19 10:43:02.193885685 -0400
>     @@ -1,4 +1,5 @@
>      Running nvme/004
>     +common/multipath-over-rdma: line 409: warning: command substitution: ignored null byte in input
>      91fdba0d-f87b-4c25-b80f-db7be1418b9e
>      uuid.91fdba0d-f87b-4c25-b80f-db7be1418b9e
>      NQN:blktests-subsystem-1 disconnected 1 controller(s)
> 
> manually to reproduce:
> Update one network interface with 15 chars:
> [blktests]# ip a s enp0s29u1u7u3c2
> 5: enp0s29u1u7u3c2: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default qlen 1000
>     link/ether f0:d4:e2:e6:e1:e3 brd ff:ff:ff:ff:ff:ff
> [blktests]# modprobe rdma_rxe
> [blktests]# echo enp0s29u1u7u3c2 >/sys/module/rdma_rxe/parameters/add
> [blktests]# cat /sys/class/infiniband/rxe0/parent
> enp0s29u1u7u3c2[blktests]# f="/sys/class/infiniband/rxe0/parent"
> [blktests]# echo $(<"$f")
> -bash: warning: command substitution: ignored null byte in input
> enp0s29u1u7u3c2
> [blktests]# echo $(tr -d '\0' <"$f")
> enp0s29u1u7u3c2
> 
> Signed-off-by: Yi Zhang <yi.zhang@redhat.com>
> ---
>  common/multipath-over-rdma | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/common/multipath-over-rdma b/common/multipath-over-rdma
> index 355b169..86e7d86 100644
> --- a/common/multipath-over-rdma
> +++ b/common/multipath-over-rdma
> @@ -406,7 +406,7 @@ has_rdma_rxe() {
>  	local f
>  
>  	for f in /sys/class/infiniband/*/parent; do
> -		if [ -e "$f" ] && [ "$(<"$f")" = "$1" ]; then
> +		if [ -e "$f" ] && [ "$(tr -d '\0' <"$f")" = "$1" ]; then
>  			return 0
>  		fi
>  	done

sysfs reads should not yield '\0' bytes. Please fix the kernel code that yields
a string including '\0' when reading from sysfs.

Thanks,

Bart.