xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

RDMA/cxgb3: When a user QP is marked in error, also mark the CQs in error

Browse Source 
The flushing of work requests for user QPs is implemented entirely in
the user mode library.  The only kernel interaction is to mark the
user QP object indicating it is in error when the QP exits RTS.  When
the user QP operations are called by the application (eg: post_send,
post_recv), the QP in error bit is checked and if set, the library
flushes the QP.  If, however, the application is not doing IO, but
rather just polling the CQ, it will never get flushed work requests.
This breaks some classes of applications.

This patch adds logic to mark user CQs in error when a QP that is bound
to the CQ is marked in error.  The library poll code can then notice
the CQ is in error and flush all the in error QPs bound to that CQ.

Design:

 - add 1 extra CQE entry to the CQ memory that will be used to indicate
   in error status.
 - return the desired CQ memory size that should be mapped by the library
 - bump the ABI since the create_cq uverbs response changes.
 - detect older libraries and reduce the mmap size accordingly.
   (The ABI bump doesn't break old libraries, since they didn't check
   the ABI field anyway)

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
This commit is contained in:
Steve Wise
2010-10-21 12:37:06 +00:00
committed by Roland Dreier
parent 252a52aa4f
commit b955150ea7
6 changed files with 71 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
drivers/infiniband/hw/cxgb3/iwch_ev.c
View File

@@ -76,6 +76,14 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp,
atomic_inc(&qhp->refcnt);
spin_unlock(&rnicp->lock);
if (qhp->attr.state == IWCH_QP_STATE_RTS) {
attrs.next_state = IWCH_QP_STATE_TERMINATE;
iwch_modify_qp(qhp->rhp, qhp, IWCH_QP_ATTR_NEXT_STATE,
&attrs, 1);
if (send_term)
iwch_post_terminate(qhp, rsp_msg);
}
event.event = ib_event;
event.device = chp->ibcq.device;
if (ib_event == IB_EVENT_CQ_ERR)
@@ -86,13 +94,7 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp,
if (qhp->ibqp.event_handler)
(*qhp->ibqp.event_handler)(&event, qhp->ibqp.qp_context);
if (qhp->attr.state == IWCH_QP_STATE_RTS) {
attrs.next_state = IWCH_QP_STATE_TERMINATE;
iwch_modify_qp(qhp->rhp, qhp, IWCH_QP_ATTR_NEXT_STATE,
&attrs, 1);
if (send_term)
iwch_post_terminate(qhp, rsp_msg);
}
(*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
if (atomic_dec_and_test(&qhp->refcnt))
wake_up(&qhp->wait);
@@ -179,7 +181,6 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb)
case TPT_ERR_BOUND:
case TPT_ERR_INVALIDATE_SHARED_MR:
case TPT_ERR_INVALIDATE_MR_WITH_MW_BOUND:
(*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
post_qp_event(rnicp, chp, rsp_msg, IB_EVENT_QP_ACCESS_ERR, 1);
break;