nvmf/ctrlr: remove assert for qpair state.
Historically, we have immediately called the qpair fini function directly from _spdk_nvmf_qpair_destroy. However, to fix a race with initiators where we can get a duplicate QID from the initiator, we will have to clear the QID bit before calling qpair_fini. Clearing the QID bit must be done on the controller thread, so we have to pass at least two messages between calling _spdk_nvmf_qpair_destroy and qpair_fini. This leaves a gap where we can poll the poll group and reap completions for the qpair before we call qpair_fini and drain all of the requests related with the qpair. Signed-off-by: Seth Howell <seth.howell@intel.com> Change-Id: I337a0608f88fdda132b8a629a508c0263d2261c5 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1881 Community-CI: Mellanox Build Bot Community-CI: Broadcom CI Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This commit is contained in:
parent
646f0665d5
commit
9d39be44c5
@ -2789,8 +2789,6 @@ spdk_nvmf_qpair_request_cleanup(struct spdk_nvmf_qpair *qpair)
|
||||
if (TAILQ_EMPTY(&qpair->outstanding)) {
|
||||
qpair->state_cb(qpair->state_cb_arg, 0);
|
||||
}
|
||||
} else {
|
||||
assert(qpair->state == SPDK_NVMF_QPAIR_ACTIVE);
|
||||
}
|
||||
}
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user