Re: [PATCH for-next 2/3] RDMA/rtrs: Fix rxe_dealloc_pd warning

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Apr 21, 2023 at 3:38 AM Zhijian Li (Fujitsu)
<lizhijian@xxxxxxxxxxx> wrote:
>
> Jinpu
>
> I updated the changes as below, and tested for thousand rounds.
>
>  From d441c0e2496c1795b5af2b6b8ae4672203d6af3c Mon Sep 17 00:00:00 2001
> From: Li Zhijian <lizhijian@xxxxxxxxxxx>
> Date: Thu, 20 Apr 2023 17:28:28 +0800
> Subject: [PATCH] RDMA/rtrs: Fix rxe_dealloc_pd warning
>
> In current design:
> 1. PD and clt_path->s.dev are shared among connections.
> 2. every con[n]'s cleanup phase will call destroy_con_cq_qp()
> 3. clt_path->s.dev will be always decreased in destroy_con_cq_qp(), and
>     when clt_path->s.dev become zero, it will destroy PD.
> 4. when con[1] failed to create, con[1] will not take clt_path->s.dev,
>     but it try to decreased clt_path->s.dev
>
> So, in case create_cm(con[0]) succeeds but create_cm(con[1])
> fails, destroy_con_cq_qp(con[1]) will be called first which will destory
> the PD while this PD is still taken by con[0].
>
> Here, we refactor the error path of create_cm() and init_conns(), so that
> we do the cleanup in the order they are created.
>
> Signed-off-by: Li Zhijian <lizhijian@xxxxxxxxxxx>
> ---
>   drivers/infiniband/ulp/rtrs/rtrs-clt.c | 47 +++++++++++---------------
>   1 file changed, 19 insertions(+), 28 deletions(-)
>
> diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.c b/drivers/infiniband/ulp/rtrs/rtrs-clt.c
> index 80abf45a197a..5faf0ecb726b 100644
> --- a/drivers/infiniband/ulp/rtrs/rtrs-clt.c
> +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.c
> @@ -2040,6 +2040,7 @@ static int rtrs_clt_rdma_cm_handler(struct rdma_cm_id *cm_id,
>         return 0;
>   }
>
> +/* The caller should the do the cleanup in case of error */
>   static int create_cm(struct rtrs_clt_con *con)
>   {
>         struct rtrs_path *s = con->c.path;
> @@ -2062,14 +2063,14 @@ static int create_cm(struct rtrs_clt_con *con)
>         err = rdma_set_reuseaddr(cm_id, 1);
>         if (err != 0) {
>                 rtrs_err(s, "Set address reuse failed, err: %d\n", err);
> -               goto destroy_cm;
> +               return err;
>         }
>         err = rdma_resolve_addr(cm_id, (struct sockaddr *)&clt_path->s.src_addr,
>                                 (struct sockaddr *)&clt_path->s.dst_addr,
>                                 RTRS_CONNECT_TIMEOUT_MS);
>         if (err) {
>                 rtrs_err(s, "Failed to resolve address, err: %d\n", err);
> -               goto destroy_cm;
> +               return err;
>         }
>         /*
>          * Combine connection status and session events. This is needed
> @@ -2084,29 +2085,17 @@ static int create_cm(struct rtrs_clt_con *con)
>                 if (err == 0)
>                         err = -ETIMEDOUT;
>                 /* Timedout or interrupted */
> -               goto errr;
> +               return err;
>         }
>         if (con->cm_err < 0) {
> -               err = con->cm_err;
> -               goto errr;
> +               return con->cm_err;
>         }
The bracket can be removed too
>         if (READ_ONCE(clt_path->state) != RTRS_CLT_CONNECTING) {
>                 /* Device removal */
> -               err = -ECONNABORTED;
> -               goto errr;
> +               return -ECONNABORTED;
>         }
same here.
>
>         return 0;
> -
> -errr:
> -       stop_cm(con);
> -       mutex_lock(&con->con_mutex);
> -       destroy_con_cq_qp(con);
> -       mutex_unlock(&con->con_mutex);
> -destroy_cm:
> -       destroy_cm(con);
> -
> -       return err;
>   }
>
>   static void rtrs_clt_path_up(struct rtrs_clt_path *clt_path)
> @@ -2334,7 +2323,7 @@ static void rtrs_clt_close_work(struct work_struct *work)
>   static int init_conns(struct rtrs_clt_path *clt_path)
>   {
>         unsigned int cid;
> -       int err;
> +       int err, i;
>
>         /*
>          * On every new session connections increase reconnect counter
> @@ -2350,10 +2339,8 @@ static int init_conns(struct rtrs_clt_path *clt_path)
>                         goto destroy;
>
>                 err = create_cm(to_clt_con(clt_path->s.con[cid]));
> -               if (err) {
> -                       destroy_con(to_clt_con(clt_path->s.con[cid]));
> +               if (err)
>                         goto destroy;
> -               }
>         }
>         err = alloc_path_reqs(clt_path);
>         if (err)
> @@ -2364,15 +2351,19 @@ static int init_conns(struct rtrs_clt_path *clt_path)
>         return 0;
>
>   destroy:
> -       while (cid--) {
> +       /* Make sure we do the cleanup in the order they are created */
> +       for (i = 0; i <= cid; i++) {
>                 struct rtrs_clt_con *con = to_clt_con(clt_path->s.con[cid]);
Yes, this line has to be adapted.
                   struct rtrs_clt_con *con = to_clt_con(clt_path->s.con[i]);
>
> -               stop_cm(con);
> -
> -               mutex_lock(&con->con_mutex);
> -               destroy_con_cq_qp(con);
> -               mutex_unlock(&con->con_mutex);
> -               destroy_cm(con);
> +               if (!con)
> +                       break;
> +               if (con->c.cm_id) {
> +                       stop_cm(con);
> +                       mutex_lock(&con->con_mutex);
> +                       destroy_con_cq_qp(con);
> +                       mutex_unlock(&con->con_mutex);
> +                       destroy_cm(con);
> +               }
>                 destroy_con(con);
>         }
>         /*
> --
> 2.29.2
>
This version looks fine. I will run some tests.

Thx!
>
>
> On 20/04/2023 10:00, Li Zhijian wrote:
> > On 19/04/2023 21:20, Jinpu Wang wrote:
> >> On Wed, Apr 19, 2023 at 11:53 AM Zhijian Li (Fujitsu)
> >> <lizhijian@xxxxxxxxxxx> wrote:
> >>>
> >>> Leon, Guoqing
> >>>
> >>>
> >>> On 18/04/2023 15:57, Leon Romanovsky wrote:
> >>>>>>> Currently, without this patch:
> >>>>>>> 1. PD and clt_path->s.dev are shared among connections.
> >>>>>>> 2. every con[n]'s cleanup phase will call destroy_con_cq_qp()
> >>>>>>> 3. clt_path->s.dev will be always decreased in destroy_con_cq_qp(), and when
> >>>>>>>        clt_path->s.dev become zero, it will destroy PD.
> >>>>>>> 4. when con[1] failed to create, con[1] will not take clt_path->s.dev, but it try to decreased clt_path->s.dev <<< it's wrong to do that.
> >>>>>> So please fix it by making sure that failure to create con[1] will
> >>>>>> release resources which were allocated. If con[1] didn't increase
> >>>>>> s.dev_ref, it shouldn't decrease it either.
> >>>>> You are right, the current patch did exactly that.
> >>>>> It introduced a con owning flag 'has_dev' to indicate whether this con has taken s.dev.
> >>>>> so that its cleanup phase will only decrease its s.dev properly.
> >>>> The has_dev is a workaround and not a solution. In proper error unwind
> >>>> sequence, you won't need extra flag.




[Index of Archives]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Photo]     [Yosemite News]     [Yosemite Photos]     [Linux Kernel]     [Linux SCSI]     [XFree86]

  Powered by Linux