Hi Zhijian, Guoqing, Leon, Bob
First of all, thanks for the patch and discussion.
On Fri, Apr 14, 2023 at 7:37 AM Zhijian Li (Fujitsu)
<lizhijian@xxxxxxxxxxx> wrote:
>
>
>
> On 14/04/2023 11:40, Guoqing Jiang wrote:
> >
> >
> > On 4/13/23 16:12, Zhijian Li (Fujitsu) wrote:
> >> On 13/04/2023 15:35, Guoqing Jiang wrote:
> >>> Hi,
> >>>
> >>> I take a closer look today.
> >>>
> >>> On 4/12/23 09:15, Zhijian Li (Fujitsu) wrote:
> >>>> On 11/04/2023 20:26, Leon Romanovsky wrote:
> >>>>> On Tue, Apr 11, 2023 at 02:43:46AM +0000, Zhijian Li (Fujitsu) wrote:
> >>>>>> On 10/04/2023 21:10, Guoqing Jiang wrote:
> >>>>>>> On 4/10/23 20:08, Leon Romanovsky wrote:
> >>>>>>>> On Mon, Apr 10, 2023 at 06:43:03AM +0000, Li Zhijian wrote:
> >>>>>>>>> The warning occurs when destroying PD whose reference count is not zero.
> >>>>>>>>>
> >>>>>>>>> Precodition: clt_path->s.con_num is 2.
> >>>>>>>>> So 2 cm connection will be created as below:
> >>>>>>>>> CPU0 CPU1
> >>>>>>>>> init_conns { |
> >>>>>>>>> create_cm() // a. con[0] created |
> >>>>>>>>> | a'. rtrs_clt_rdma_cm_handler() {
> >>>>>>>>> | rtrs_rdma_addr_resolved()
> >>>>>>>>> | create_con_cq_qp(con); << con[0]
> >>>>>>>>> | }
> >>>>>>>>> | in this moment, refcnt of PD was increased to 2+
> >>> What do you mean "refcnt of PD"? usecnt in struct ib_pd or dev_ref.
> >> I mean usecnt in struct ib_pd
> >>
> >>
> >>
> >>>>>>>>> |
> >>>>>>>>> create_cm() // b. cid = 1, failed |
> >>>>>>>>> destroy_con_cq_qp() |
> >>>>>>>>> rtrs_ib_dev_put() |
> >>>>>>>>> dev_free() |
> >>>>>>>>> ib_dealloc_pd(dev->ib_pd) << PD |
> >>>>>>>>> is destroyed, but refcnt is |
> >>>>>>>>> still greater than 0 |
> >>> Assuming you mean "pd->usecnt". We only allocate pd in con[0] by rtrs_ib_dev_find_or_add,
> >>> if con[1] failed to create cm, then alloc_path_reqs -> ib_alloc_mr -> atomic_inc(&pd->usecnt)
> >
> > The above can't be invoked, right?
> >
> >>> can't be triggered. Is there other places could increase the refcnt?
> >> Yes, when create a qp, it will also associate to this PD, that also mean refcnt of PD will be increased.
> >>
> >> When con[0](create_con_cq_qp) succeeded, refcnt of PD will be 2. and then when con[1] failed, since
> >> QP didn't create, refcnt of PD is still 2. con[1]'s cleanup will destroy the PD(ib_dealloc_pd) since dev_ref = 1, after that its
> >> refcnt is still 1.
> >
> > I can see the path increase usecnt to 1.
> >
> > rtrs_cq_qp_create -> create_qp
> > -> rdma_create_qp
> > -> ib_create_qp
> > -> create_qp
> > -> ib_qp_usecnt_inc which increases pd->usecnt
> >
> > Where is another place to increase usecnt to 2?
>
> It should be
> ib_create_qp ...
> -> rxe_create_qp
> -> rxe_qp_from_init
> -> rxe_get(pd) <<< pd's refcnt will be increased.
IIUC, this problem is rxe specific, because rxe manipulate refcnt
itself? I checked mlx5/mlx4 they do not change the refcnt of pd when
create_kernel_qp.
So question is then if the bug is on rxe side or rtrs side?
Zhijian how do you reproduce the warning? do you inject error explictly?
Regards!
>
>
> >
> >>> Then what is the appropriate time to call destroy_con_cq_qp for this scenario?
> >>> Otherwise there could be memory leak.
> >> we must ensure QP in con[0] is closed before destroying the PD.
> >> Currently destroy_con_cq_qp() subroutine will close the opened QP first.
> >
> > Let me try another way, with below change, rtrs_ib_dev_put can't be called
> > from destroy_con_cq_qp, right?
>
> Not really, con[0]->has_dev is true, so con[0]'s cleanup will call rtrs_ib_dev_put()
>
> Without this patch, when con[1] failed, con[1]'s cleanup will be called first. then call con[0]'s cleanup.
> After this change, con[1]'s cleanup will not call rtrs_ib_dev_put, but it will be called the later con[0]'s cleanup.
>
>
> Thanks
> Zhijian
>
> >
> > + if (!con->has_dev)
> > + return;
> > if (clt_path->s.dev_ref && !--clt_path->s.dev_ref) {
> > rtrs_ib_dev_put(clt_path->s.dev);
> > clt_path->s.dev = NULL;
> >
> > Then when will you dealloc pd and free rtrs_ib_dev?
> >
> > Thanks,
> > Guoqing
