IB: Return "maybe missed event" hint from ib_req_notify_cq()
The semantics defined by the InfiniBand specification say that
completion events are only generated when a completions is added to a
completion queue (CQ) after completion notification is requested. In
other words, this means that the following race is possible:
while (CQ is not empty)
ib_poll_cq(CQ);
// new completion is added after while loop is exited
ib_req_notify_cq(CQ);
// no event is generated for the existing completion
To close this race, the IB spec recommends doing another poll of the
CQ after requesting notification.
However, it is not always possible to arrange code this way (for
example, we have found that NAPI for IPoIB cannot poll after
requesting notification). Also, some hardware (eg Mellanox HCAs)
actually will generate an event for completions added before the call
to ib_req_notify_cq() -- which is allowed by the spec, since there's
no way for any upper-layer consumer to know exactly when a completion
was really added -- so the extra poll of the CQ is just a waste.
Motivated by this, we add a new flag "IB_CQ_REPORT_MISSED_EVENTS" for
ib_req_notify_cq() so that it can return a hint about whether the a
completion may have been added before the request for notification.
The return value of ib_req_notify_cq() is extended so:
< 0 means an error occurred while requesting notification
== 0 means notification was requested successfully, and if
IB_CQ_REPORT_MISSED_EVENTS was passed in, then no
events were missed and it is safe to wait for another
event.
> 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was
passed in. It means that the consumer must poll the
CQ again to make sure it is empty to avoid the race
described above.
We add a flag to enable this behavior rather than turning it on
unconditionally, because checking for missed events may incur
significant overhead for some low-level drivers, and consumers that
don't care about the results of this test shouldn't be forced to pay
for the test.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 17cc309..5342ac6 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -431,9 +431,11 @@
u8 port_num; /* valid only for DR SMPs on switches */
};
-enum ib_cq_notify {
- IB_CQ_SOLICITED,
- IB_CQ_NEXT_COMP
+enum ib_cq_notify_flags {
+ IB_CQ_SOLICITED = 1 << 0,
+ IB_CQ_NEXT_COMP = 1 << 1,
+ IB_CQ_SOLICITED_MASK = IB_CQ_SOLICITED | IB_CQ_NEXT_COMP,
+ IB_CQ_REPORT_MISSED_EVENTS = 1 << 2,
};
enum ib_srq_attr_mask {
@@ -990,7 +992,7 @@
struct ib_wc *wc);
int (*peek_cq)(struct ib_cq *cq, int wc_cnt);
int (*req_notify_cq)(struct ib_cq *cq,
- enum ib_cq_notify cq_notify);
+ enum ib_cq_notify_flags flags);
int (*req_ncomp_notif)(struct ib_cq *cq,
int wc_cnt);
struct ib_mr * (*get_dma_mr)(struct ib_pd *pd,
@@ -1419,14 +1421,34 @@
/**
* ib_req_notify_cq - Request completion notification on a CQ.
* @cq: The CQ to generate an event for.
- * @cq_notify: If set to %IB_CQ_SOLICITED, completion notification will
- * occur on the next solicited event. If set to %IB_CQ_NEXT_COMP,
- * notification will occur on the next completion.
+ * @flags:
+ * Must contain exactly one of %IB_CQ_SOLICITED or %IB_CQ_NEXT_COMP
+ * to request an event on the next solicited event or next work
+ * completion at any type, respectively. %IB_CQ_REPORT_MISSED_EVENTS
+ * may also be |ed in to request a hint about missed events, as
+ * described below.
+ *
+ * Return Value:
+ * < 0 means an error occurred while requesting notification
+ * == 0 means notification was requested successfully, and if
+ * IB_CQ_REPORT_MISSED_EVENTS was passed in, then no events
+ * were missed and it is safe to wait for another event. In
+ * this case is it guaranteed that any work completions added
+ * to the CQ since the last CQ poll will trigger a completion
+ * notification event.
+ * > 0 is only returned if IB_CQ_REPORT_MISSED_EVENTS was passed
+ * in. It means that the consumer must poll the CQ again to
+ * make sure it is empty to avoid missing an event because of a
+ * race between requesting notification and an entry being
+ * added to the CQ. This return value means it is possible
+ * (but not guaranteed) that a work completion has been added
+ * to the CQ since the last poll without triggering a
+ * completion notification event.
*/
static inline int ib_req_notify_cq(struct ib_cq *cq,
- enum ib_cq_notify cq_notify)
+ enum ib_cq_notify_flags flags)
{
- return cq->device->req_notify_cq(cq, cq_notify);
+ return cq->device->req_notify_cq(cq, flags);
}
/**