IB/qib: Avoid returning EBUSY from MR deregister

A timing issue can occur where qib_mr_dereg can return -EBUSY if the
MR use count is not zero.

This can occur if the MR is de-registered while RDMA read response
packets are being progressed from the SDMA ring.  The suspicion is
that the peer sent an RDMA read request, which has already been copied
across to the peer.  The peer sees the completion of his request and
then communicates to the responder that the MR is not needed any
longer.  The responder tries to de-register the MR, catching some
responses remaining in the SDMA ring holding the MR use count.

The code now uses a get/put paradigm to track MR use counts and
coordinates with the MR de-registration process using a completion
when the count has reached zero.  A timeout on the delay is in place
to catch other EBUSY issues.

The reference count protocol is as follows:
- The return to the user counts as 1
- A reference from the lk_table or the qib_ibdev counts as 1.
- Transient I/O operations increase/decrease as necessary

A lot of code duplication has been folded into the new routines
init_qib_mregion() and deinit_qib_mregion().  Additionally, explicit
initialization of fields to zero is now handled by kzalloc().

Also, duplicated code 'while.*num_sge' that decrements reference
counts have been consolidated in qib_put_ss().

Reviewed-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

1 files changed, 26 insertions, 2 deletions

diff --git a/drivers/infiniband/hw/qib/qib_verbs.h b/drivers/infiniband/hw/qib/qib_verbs.h
index 487606024659..4a2277bc059e 100644
--- a/drivers/infiniband/hw/qib/qib_verbs.h
+++ b/drivers/infiniband/hw/qib/qib_verbs.h
@@ -41,6 +41,7 @@
 #include <linux/interrupt.h>
 #include <linux/kref.h>
 #include <linux/workqueue.h>
+#include <linux/completion.h>
 #include <rdma/ib_pack.h>
 #include <rdma/ib_user_verbs.h>
 
@@ -302,6 +303,8 @@ struct qib_mregion {
         u32 max_segs;           /* number of qib_segs in all the arrays */
         u32 mapsz;              /* size of the map array */
         u8  page_shift;         /* 0 - non unform/non powerof2 sizes */
+        u8  lkey_published;     /* in global table */
+        struct completion comp; /* complete when refcount goes to zero */
         atomic_t refcount;
         struct qib_segarray *map[0];    /* the segments */
 };
@@ -944,9 +947,9 @@ int qib_post_ud_send(struct qib_qp *qp, struct ib_send_wr *wr);
 void qib_ud_rcv(struct qib_ibport *ibp, struct qib_ib_header *hdr,
                 int has_grh, void *data, u32 tlen, struct qib_qp *qp);
 
-int qib_alloc_lkey(struct qib_lkey_table *rkt, struct qib_mregion *mr);
+int qib_alloc_lkey(struct qib_mregion *mr, int dma_region);
 
-int qib_free_lkey(struct qib_ibdev *dev, struct qib_mregion *mr);
+void qib_free_lkey(struct qib_mregion *mr);
 
 int qib_lkey_ok(struct qib_lkey_table *rkt, struct qib_pd *pd,
                 struct qib_sge *isge, struct ib_sge *sge, int acc);
@@ -1014,6 +1017,27 @@ int qib_unmap_fmr(struct list_head *fmr_list);
 
 int qib_dealloc_fmr(struct ib_fmr *ibfmr);
 
+static inline void qib_get_mr(struct qib_mregion *mr)
+{
+        atomic_inc(&mr->refcount);
+}
+
+static inline void qib_put_mr(struct qib_mregion *mr)
+{
+        if (unlikely(atomic_dec_and_test(&mr->refcount)))
+                complete(&mr->comp);
+}
+
+static inline void qib_put_ss(struct qib_sge_state *ss)
+{
+        while (ss->num_sge) {
+                qib_put_mr(ss->sge.mr);
+                if (--ss->num_sge)
+                        ss->sge = *ss->sg_list++;
+        }
+}
+
+
 void qib_release_mmap_info(struct kref *ref);
 
 struct qib_mmap_info *qib_create_mmap_info(struct qib_ibdev *dev, u32 size,