RDMA/rxe: Add rxe_is_fenced() subroutine
authorBob Pearson <rpearsonhpe@gmail.com>
Thu, 30 Jun 2022 19:04:18 +0000 (14:04 -0500)
committerJason Gunthorpe <jgg@nvidia.com>
Fri, 22 Jul 2022 20:43:00 +0000 (17:43 -0300)
commit930119a1720075d15e4c1e478b2b9412cd9eb6ad
tree1845fe390ebac33807690f2c00fa77a8cf4b0fe2
parent174e7b137042f19b5ce88beb4fc0ff4ec6b0c72a
RDMA/rxe: Add rxe_is_fenced() subroutine

The code thc that decides whether to defer execution of a wqe in
rxe_requester.c is isolated into a subroutine rxe_is_fenced() and removed
from the call to req_next_wqe(). The condition whether a wqe should be
fenced is changed to comply with the IBA. Currently an operation is fenced
if the fence bit is set in the wqe flags and the last wqe has not
completed. For normal operations the IBA actually only requires that the
last read or atomic operation is complete.

Link: https://lore.kernel.org/r/20220630190425.2251-2-rpearsonhpe@gmail.com
Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Reviewed-by: Li Zhijian <lizhijian@fujitsu.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
drivers/infiniband/sw/rxe/rxe_req.c