diff options
author | Dakshaja Uppalapati <dakshaja@chelsio.com> | 2021-08-05 18:13:32 +0530 |
---|---|---|
committer | Jason Gunthorpe <jgg@nvidia.com> | 2021-08-05 22:33:00 -0300 |
commit | 2638a32348bbb1c384dbbd515fd2b12c155f0188 (patch) | |
tree | 9fb7afbc32e8c6172a2cac36a79c88e51b5aefb1 /drivers/infiniband/hw/cxgb4/ev.c | |
parent | RDMA/hns: Fix the double unlock problem of poll_sem (diff) | |
download | linux-dev-2638a32348bbb1c384dbbd515fd2b12c155f0188.tar.xz linux-dev-2638a32348bbb1c384dbbd515fd2b12c155f0188.zip |
RDMA/iw_cxgb4: Fix refcount underflow while destroying cqs.
Previous atomic increment/decrement logic expects the atomic count to be
'0' after the final decrement.
Replacing atomic count with refcount does not allow that, as
refcount_dec() considers count of 1 as underflow and triggers a kernel
splat.
Fix the current refcount logic by using the usual pattern of decrementing
the refcount and test if it is '0' on the final deref in
c4iw_destroy_cq(). Use wait_for_completion() instead of wait_event().
Fixes: 7183451f846d ("RDMA/cxgb4: Use refcount_t instead of atomic_t for reference counting")
Link: https://lore.kernel.org/r/1628167412-12114-1-git-send-email-dakshaja@chelsio.com
Signed-off-by: Dakshaja Uppalapati <dakshaja@chelsio.com>
Reviewed-by: Potnuri Bharat Teja <bharat@chelsio.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Diffstat (limited to 'drivers/infiniband/hw/cxgb4/ev.c')
-rw-r--r-- | drivers/infiniband/hw/cxgb4/ev.c | 6 |
1 files changed, 2 insertions, 4 deletions
diff --git a/drivers/infiniband/hw/cxgb4/ev.c b/drivers/infiniband/hw/cxgb4/ev.c index 7798d090888b..34211a533d5c 100644 --- a/drivers/infiniband/hw/cxgb4/ev.c +++ b/drivers/infiniband/hw/cxgb4/ev.c @@ -213,8 +213,7 @@ void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe) break; } done: - if (refcount_dec_and_test(&chp->refcnt)) - wake_up(&chp->wait); + c4iw_cq_rem_ref(chp); c4iw_qp_rem_ref(&qhp->ibqp); out: return; @@ -234,8 +233,7 @@ int c4iw_ev_handler(struct c4iw_dev *dev, u32 qid) spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); spin_unlock_irqrestore(&chp->comp_handler_lock, flag); - if (refcount_dec_and_test(&chp->refcnt)) - wake_up(&chp->wait); + c4iw_cq_rem_ref(chp); } else { pr_debug("unknown cqid 0x%x\n", qid); xa_unlock_irqrestore(&dev->cqs, flag); |