On Thu, Jun 25, 2020 at 10:11:07AM -0700, Divya Indi wrote:
> Hi Leon,
>
> Please find my comments inline -
>
> On 6/25/20 3:09 AM, Leon Romanovsky wrote:
> > On Tue, Jun 23, 2020 at 07:13:09PM -0700, Divya Indi wrote:
> >> Commit 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending")'
> >> -
> >> 1. Adds the query to the request list before ib_nl_snd_msg.
> >> 2. Moves ib_nl_send_msg out of spinlock, hence safe to use gfp_mask as is.
> >>
> >> However, if there is a delay in sending out the request (For
> >> eg: Delay due to low memory situation) the timer to handle request timeout
> >> might kick in before the request is sent out to ibacm via netlink.
> >> ib_nl_request_timeout may release the query causing a use after free situation
> >> while accessing the query in ib_nl_send_msg.
> >>
> >> Call Trace for the above race:
> >>
> >> [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core]
> >> [<ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa]
> >> [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm]
> >> [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm]
> >> [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850
> >> [rds_rdma]
> >> [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850
> >> [rds_rdma]
> >> [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma]
> >> [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm]
> >> [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr]
> >> [<ffffffff810a02f9>] process_one_work+0x169/0x4a0
> >> [<ffffffff810a0b2b>] worker_thread+0x5b/0x560
> >> [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50
> >> [<ffffffff810a68fb>] kthread+0xcb/0xf0
> >> [<ffffffff816ec49a>] ? __schedule+0x24a/0x810
> >> [<ffffffff816ec49a>] ? __schedule+0x24a/0x810
> >> [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180
> >> [<ffffffff816f25a7>] ret_from_fork+0x47/0x90
> >> [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180
> >> ....
> >> RIP [<ffffffffa03296cd>] send_mad+0x33d/0x5d0 [ib_sa]
> >>
> >> To resolve the above issue -
> >> 1. Add the req to the request list only after the request has been sent out.
> >> 2. To handle the race where response comes in before adding request to
> >> the request list, send(rdma_nl_multicast) and add to list while holding the
> >> spinlock - request_lock.
> >> 3. Use non blocking memory allocation flags for rdma_nl_multicast since it is
> >> called while holding a spinlock.
> >>
> >> Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list
> >> before sending")
> >>
> >> Signed-off-by: Divya Indi <divya.indi@xxxxxxxxxx>
> >> ---
> >> v1:
> >> - Use flag IB_SA_NL_QUERY_SENT to prevent the use-after-free.
> >>
> >> v2:
> >> - Use atomic bit ops for setting and testing IB_SA_NL_QUERY_SENT.
> >> - Rewording and adding comments.
> >>
> >> v3:
> >> - Change approach and remove usage of IB_SA_NL_QUERY_SENT.
> >> - Add req to request list only after the request has been sent out.
> >> - Send and add to list while holding the spinlock (request_lock).
> >> - Overide gfp_mask and use GFP_NOWAIT for rdma_nl_multicast since we
> >> need non blocking memory allocation while holding spinlock.
> >>
> >> v4:
> >> - Formatting changes.
> >> - Use GFP_NOWAIT conditionally - Only when GFP_ATOMIC is not provided by caller.
> >> ---
> >> drivers/infiniband/core/sa_query.c | 41 ++++++++++++++++++++++----------------
> >> 1 file changed, 24 insertions(+), 17 deletions(-)
> >>
> >> diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c
> >> index 74e0058..9066d48 100644
> >> --- a/drivers/infiniband/core/sa_query.c
> >> +++ b/drivers/infiniband/core/sa_query.c
> >> @@ -836,6 +836,10 @@ static int ib_nl_send_msg(struct ib_sa_query *query, gfp_t gfp_mask)
> >> void *data;
> >> struct ib_sa_mad *mad;
> >> int len;
> >> + unsigned long flags;
> >> + unsigned long delay;
> >> + gfp_t gfp_flag;
> >> + int ret;
> >>
> >> mad = query->mad_buf->mad;
> >> len = ib_nl_get_path_rec_attrs_len(mad->sa_hdr.comp_mask);
> >> @@ -860,36 +864,39 @@ static int ib_nl_send_msg(struct ib_sa_query *query, gfp_t gfp_mask)
> >> /* Repair the nlmsg header length */
> >> nlmsg_end(skb, nlh);
> >>
> >> - return rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, gfp_mask);
> >> -}
> >> + gfp_flag = ((gfp_mask & GFP_ATOMIC) == GFP_ATOMIC) ? GFP_ATOMIC :
> >> + GFP_NOWAIT;
> > I would say that the better way will be to write something like this:
> > gfp_flag |= GFP_NOWAIT;
>
> You mean gfp_flag = gfp_mask|GFP_NOWAIT? [We dont want to modify the gfp_mask sent by caller]
>
> #define GFP_ATOMIC (__GFP_HIGH|__GFP_ATOMIC|__GFP_KSWAPD_RECLAIM)
> #define GFP_KERNEL (__GFP_RECLAIM | __GFP_IO | __GFP_FS)
> #define GFP_NOWAIT (__GFP_KSWAPD_RECLAIM)
>
> If a caller passes GFP_KERNEL, "gfp_mask|GFP_NOWAIT" will still have __GFP_RECLAIM,
> __GFP_IO and __GFP_FS set which is not suitable for using under spinlock.
Ahh, sorry I completely forgot about spinlock part.
Thanks
>
> Thanks,
> Divya
>
> >
> > Thanks
