Re: [PATCH blktests V3] tests/srp: fix module loading issue during srp tests

[Yi Zhang <yi.zhang@redhat.com>]

On Mon, Oct 25, 2021 at 5:22 AM Bart Van Assche <bvanassche@acm.org> wrote:
>
> On 10/24/21 05:42, Yi Zhang wrote:
> > The ib_isert/ib_srpt modules will be automatically loaded after the first
> >   time rdma_rxe/siw setup, which will lead srp tests fail.
> >
> > $ modprobe rdma_rxe
> > $ echo eno1 >/sys/module/rdma_rxe/parameters/add
> > $ lsmod | grep -E "ib_srpt|iscsi_target_mod|ib_isert"
> > ib_srpt               167936  0
> > ib_isert              139264  0
> > iscsi_target_mod      843776  1 ib_isert
> > target_core_mod      1069056  3 iscsi_target_mod,ib_srpt,ib_isert
> > rdma_cm               315392  5 rpcrdma,ib_srpt,ib_iser,ib_isert,rdma_ucm
> > ib_cm                 344064  2 rdma_cm,ib_srpt
> > ib_core              1101824  10 rdma_cm,rdma_rxe,rpcrdma,ib_srpt,iw_cm,ib_iser,ib_isert,rdma_ucm,ib_uverbs,ib_cm
> >
> > $ ./check srp/001
> > srp/001 (Create and remove LUNs)                             [failed]
> >      runtime    ...  3.675s
> >      --- tests/srp/001.out    2021-10-13 01:18:50.846740093 -0400
> >      +++ /root/blktests/results/nodev/srp/001.out.bad 2021-10-14 03:24:18.593852208 -0400
> >      @@ -1,3 +1 @@
> >      -Configured SRP target driver
> >      -count_luns(): 3 <> 3
> >      -Passed
> >      +insmod: ERROR: could not insert module /lib/modules/5.15.0-rc5.fix+/kernel/drivers/infiniband/ulp/srpt/ib_srpt.ko: File exists
> > modprobe: FATAL: Module iscsi_target_mod is in use.
> >
> > Signed-off-by: Yi Zhang <yi.zhang@redhat.com>
> > ---
> >   tests/srp/rc | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/tests/srp/rc b/tests/srp/rc
> > index 7239d87..b3dfd4d 100755
> > --- a/tests/srp/rc
> > +++ b/tests/srp/rc
> > @@ -497,7 +497,7 @@ start_lio_srpt() {
> >       if modinfo ib_srpt | grep -q '^parm:[[:blank:]]*rdma_cm_port:'; then
> >               opts+=("rdma_cm_port=${srp_rdma_cm_port}")
> >       fi
> > -     insmod "/lib/modules/$(uname -r)/kernel/drivers/infiniband/ulp/srpt/ib_srpt."* "${opts[@]}" || return $?
> > +     unload_module ib_srpt && modprobe ib_srpt "${opts[@]}" || return $?
> >       i=0
>
> The "&&" above seems wrong to me. It is not guaranteed that the ib_srpt
> kernel mode has already been loaded before this code runs. I propose to
> use the following code instead:
>
>         unload_module ib_srpt
>         modprobe ib_srpt "${opts[@]}" || return $?

Thanks Bart, send v4 for this change.

>
> Thanks,
>
> Bart.
>


-- 
Best Regards,
  Yi Zhang