IB/mlx5: Reset flow support for IB kernel ULPs
authorMaor Gottlieb <maorg@mellanox.com>
Fri, 17 Jun 2016 12:01:38 +0000 (15:01 +0300)
committerDoug Ledford <dledford@redhat.com>
Thu, 23 Jun 2016 15:02:45 +0000 (11:02 -0400)
commit89ea94a7b6c40eb423c144aef1caceebaff79c8d
tree2831167a7aecf1b24275d3aca92bc1d3fadcc107
parent7c2344c3bbf97eb5dfa732d5098285d15d3bf9bf
IB/mlx5: Reset flow support for IB kernel ULPs

The driver exposes interfaces that directly relate to HW state.
Upon fatal error, consumers of these interfaces (ULPs) that rely
on completion of all their posted work-request could hang, thereby
introducing dependencies in shutdown order. To prevent this from
happening, we manage the relevant resources (CQs, QPs) that are used
by the device. Upon a fatal error, we now generate simulated
completions for outstanding WQEs that were not completed at the
time the HW was reset.

It includes invoking the completion event handler for all involved
CQs so that the ULPs will poll those CQs. When polled we return
simulated CQEs with IB_WC_WR_FLUSH_ERR return code enabling ULPs
to clean up their  resources and not wait forever for completions
upon receiving remove_one.

The above change requires an extra check in the data path to make
sure that when device is in error state, the simulated CQEs will
be returned and no further WQEs will be posted.

Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>
drivers/infiniband/hw/mlx5/cq.c
drivers/infiniband/hw/mlx5/main.c
drivers/infiniband/hw/mlx5/mlx5_ib.h
drivers/infiniband/hw/mlx5/mr.c
drivers/infiniband/hw/mlx5/qp.c
drivers/infiniband/hw/mlx5/srq.c
include/linux/mlx5/cq.h