Skip to content

rdma/siw: fix a lot of deadlocks and use after free bugs

Stefan Metzmacher linux-rdma 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest31 patches 25 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/31rdma/siw: fix warning in siw_proc_send()drivers/infiniband/sw/siw/siw_qp_rx.c+1 -11 reply
2/31rdma/siw: call smp_mb() after mem->stag_valid = 0 in siw_invalidate_stag() toodrivers/infiniband/sw/siw/siw_mem.c+2 -01 reply
3/31rdma/siw: remove superfluous siw_cep_put() from siw_connect() error pathdrivers/infiniband/sw/siw/siw_cm.c+0 -11 reply
4/31rdma/siw: let siw_accept() deferr RDMA_MODE until EVENT_ESTABLISHEDdrivers/infiniband/sw/siw/siw_cm.c+4 -31 reply
5/31rdma/siw: make use of kernel_{bind,connect,listen}()drivers/infiniband/sw/siw/siw_cm.c+7 -71 reply
6/31rdma/siw: make siw_cm_upcall() a noop without valid 'id'drivers/infiniband/sw/siw/siw_cm.c+3 -01 reply
7/31rdma/siw: split out a __siw_cep_terminate_upcall() functiondrivers/infiniband/sw/siw/siw_cm.c+32 -211 reply
8/31rdma/siw: use __siw_cep_terminate_upcall() for indirect SIW_CM_WORK_CLOSE_LLPdrivers/infiniband/sw/siw/siw_cm.c+1 -51 reply
9/31rdma/siw: use __siw_cep_terminate_upcall() for SIW_CM_WORK_PEER_CLOSEdrivers/infiniband/sw/siw/siw_cm.c+50 -391 reply
10/31rdma/siw: use __siw_cep_terminate_upcall() for SIW_CM_WORK_MPATIMEOUTdrivers/infiniband/sw/siw/siw_cm.c+8 -231 reply
11/31rdma/siw: introduce SIW_EPSTATE_ACCEPTING/REJECTING for rdma_accept/rdma_rejectdrivers/infiniband/sw/siw/siw_cm.c drivers/infiniband/sw/siw/siw_cm.h+31 -21 reply
12/31rdma/siw: add some debugging of state and sk_state to the teardown processdrivers/infiniband/sw/siw/siw_cm.c+14 -11 reply
13/31rdma/siw: handle SIW_EPSTATE_CONNECTING in __siw_cep_terminate_upcall()drivers/infiniband/sw/siw/siw_cm.c+8 -10 replies
14/31rdma/siw: let siw_connect() set AWAIT_MPAREP before siw_send_mpareqrep()drivers/infiniband/sw/siw/siw_cm.c+2 -20 replies
15/31rdma/siw: create a temporary copy of private datadrivers/infiniband/sw/siw/siw_cm.c+14 -21 reply
16/31rdma/siw: use error and out logic at the end of siw_connect()drivers/infiniband/sw/siw/siw_cm.c+12 -71 reply
17/31rdma/siw: start mpa timer before calling siw_send_mpareqrep()drivers/infiniband/sw/siw/siw_cm.c+7 -51 reply
18/31rdma/siw: call the blocking kernel_bindconnect() just before siw_send_mpareqrep()drivers/infiniband/sw/siw/siw_cm.c+13 -120 replies
19/31rdma/siw: split out a __siw_cep_close() functiondrivers/infiniband/sw/siw/siw_cm.c+33 -151 reply
20/31rdma/siw: implement non-blocking connect.drivers/infiniband/sw/siw/siw_cm.c drivers/infiniband/sw/siw/siw_cm.h+85 -301 reply
21/31rdma/siw: let siw_listen_address() call siw_cep_alloc() firstdrivers/infiniband/sw/siw/siw_cm.c+19 -200 replies
22/31rdma/siw: let siw_listen_address() call siw_cep_set_inuse() earlydrivers/infiniband/sw/siw/siw_cm.c+3 -21 reply
23/31rdma/siw: make use of __siw_cep_close() in siw_accept()drivers/infiniband/sw/siw/siw_cm.c+1 -171 reply
24/31rdma/siw: do the full disassociation of cep and qp in siw_qp_llp_close()drivers/infiniband/sw/siw/siw_cm.c drivers/infiniband/sw/siw/siw_qp.c+3 -21 reply
25/31rdma/siw: fix double siw_cep_put() in siw_cm_work_handler()drivers/infiniband/sw/siw/siw_cm.c+0 -11 reply
26/31rdma/siw: make use of __siw_cep_close() in siw_cm_work_handler()drivers/infiniband/sw/siw/siw_cm.c+1 -90 replies
27/31rdma/siw: fix the "close" logic in siw_qp_cm_drop()drivers/infiniband/sw/siw/siw_cm.c+4 -21 reply
28/31rdma/siw: make use of __siw_cep_close() in siw_qp_cm_drop()drivers/infiniband/sw/siw/siw_cm.c+1 -220 replies
29/31rdma/siw: make use of __siw_cep_close() in siw_reject()drivers/infiniband/sw/siw/siw_cm.c+1 -50 replies
30/31rdma/siw: make use of __siw_cep_close() in siw_listen_address()drivers/infiniband/sw/siw/siw_cm.c+1 -90 replies
31/31rdma/siw: make use of __siw_cep_close() in siw_drop_listeners()drivers/infiniband/sw/siw/siw_cm.c+1 -110 replies

Reviews trailers per patch

8 of 31 patches have a Reviewed-by or Acked-by.

PatchBernard Metzler
1/31
2/31R-b
3/31R-b
4/31R-b
5/31R-b
6/31
7/31
8/31
9/31
10/31
11/31
12/31
13/31
14/31
15/31
16/31R-b
17/31R-b
18/31
19/31
20/31
21/31
22/31R-b
23/31R-b
24/31
25/31
26/31
27/31
28/31
29/31
30/31
31/31

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 00/31] rdma/siw: fix a lot of deadlocks and use after free bugs

Hi Bernard,

while testing with my smbdirect driver I hit a lot of bugs in the siw.ko driver. They all cause problems where the siw driver was not able to unload anymore and I had to reboot the machine.

I implemented:
- a non blocking connect
- fixed a lot of bugs where siw_cep_put() was called too often.
- fixed bugs where siw_cm_upcall() confused the core IWCM logic

I have some more changes to follow, but I wanted to send them finally out after having them one and a half year sitting in some private branch...

Stefan Metzmacher (31):
rdma/siw: fix warning in siw_proc_send()
rdma/siw: call smp_mb() after mem->stag_valid = 0 in
siw_invalidate_stag() too
rdma/siw: remove superfluous siw_cep_put() from siw_connect() error
path
rdma/siw: let siw_accept() deferr RDMA_MODE until EVENT_ESTABLISHED rdma/siw: make use of kernel_{bind,connect,listen}() rdma/siw: make siw_cm_upcall() a noop without valid 'id' rdma/siw: split out a __siw_cep_terminate_upcall() function rdma/siw: use __siw_cep_terminate_upcall() for indirect
SIW_CM_WORK_CLOSE_LLP
rdma/siw: use __siw_cep_terminate_upcall() for SIW_CM_WORK_PEER_CLOSE rdma/siw: use __siw_cep_terminate_upcall() for SIW_CM_WORK_MPATIMEOUT rdma/siw: introduce SIW_EPSTATE_ACCEPTING/REJECTING for
rdma_accept/rdma_reject
rdma/siw: add some debugging of state and sk_state to the teardown
process
rdma/siw: handle SIW_EPSTATE_CONNECTING in
__siw_cep_terminate_upcall()
rdma/siw: let siw_connect() set AWAIT_MPAREP before
siw_send_mpareqrep()
rdma/siw: create a temporary copy of private data rdma/siw: use error and out logic at the end of siw_connect() rdma/siw: start mpa timer before calling siw_send_mpareqrep() rdma/siw: call the blocking kernel_bindconnect() just before
siw_send_mpareqrep()
rdma/siw: split out a __siw_cep_close() function
rdma/siw: implement non-blocking connect.
rdma/siw: let siw_listen_address() call siw_cep_alloc() first rdma/siw: let siw_listen_address() call siw_cep_set_inuse() early rdma/siw: make use of __siw_cep_close() in siw_accept() rdma/siw: do the full disassociation of cep and qp in

    siw_qp_llp_close()

rdma/siw: fix double siw_cep_put() in siw_cm_work_handler() rdma/siw: make use of __siw_cep_close() in siw_cm_work_handler() rdma/siw: fix the "close" logic in siw_qp_cm_drop() rdma/siw: make use of __siw_cep_close() in siw_qp_cm_drop() rdma/siw: make use of __siw_cep_close() in siw_reject() rdma/siw: make use of __siw_cep_close() in siw_listen_address() rdma/siw: make use of __siw_cep_close() in siw_drop_listeners()

 drivers/infiniband/sw/siw/siw_cm.c    | 537 +++++++++++++++-----------
 drivers/infiniband/sw/siw/siw_cm.h    |   3 +
 drivers/infiniband/sw/siw/siw_mem.c   |   2 +
 drivers/infiniband/sw/siw/siw_qp.c    |   3 +
 drivers/infiniband/sw/siw/siw_qp_rx.c |   2 +-
 5 files changed, 316 insertions(+), 231 deletions(-)
Signature · 1 line
2.25.1