Merge branch 'k.o/for-rc' into k.o/wip/dl-for-next

Several items of conflict have arisen between the RDMA stack's for-rc
branch and upcoming for-next work:

9fd4350ba895 ("IB/rxe: avoid double kfree_skb") directly conflicts with
2e47350789eb ("IB/rxe: optimize the function duplicate_request")

Patches already submitted by Intel for the hfi1 driver will fail to
apply cleanly without this merge

Other people on the mailing list have notified that their upcoming
patches also fail to apply cleanly without this merge

Signed-off-by: Doug Ledford <dledford@redhat.com>

5 files changed, 49 insertions, 7 deletions

diff --git a/drivers/infiniband/hw/cxgb4/cq.c b/drivers/infiniband/hw/cxgb4/cq.c
index 6f2b26126c64..2be2e1ac1b5f 100644
--- a/drivers/infiniband/hw/cxgb4/cq.c
+++ b/drivers/infiniband/hw/cxgb4/cq.c
@@ -315,7 +315,7 @@ static void advance_oldest_read(struct t4_wq *wq)
  * Deal with out-of-order and/or completions that complete
  * prior unsignalled WRs.
  */
-void c4iw_flush_hw_cq(struct c4iw_cq *chp)
+void c4iw_flush_hw_cq(struct c4iw_cq *chp, struct c4iw_qp *flush_qhp)
 {
 	struct t4_cqe *hw_cqe, *swcqe, read_cqe;
 	struct c4iw_qp *qhp;
@@ -339,6 +339,13 @@ void c4iw_flush_hw_cq(struct c4iw_cq *chp)
 		if (qhp == NULL)
 			goto next_cqe;
 
+		if (flush_qhp != qhp) {
+			spin_lock(&qhp->lock);
+
+			if (qhp->wq.flushed == 1)
+				goto next_cqe;
+		}
+
 		if (CQE_OPCODE(hw_cqe) == FW_RI_TERMINATE)
 			goto next_cqe;
 
@@ -390,6 +397,8 @@ void c4iw_flush_hw_cq(struct c4iw_cq *chp)
 next_cqe:
 		t4_hwcq_consume(&chp->cq);
 		ret = t4_next_hw_cqe(&chp->cq, &hw_cqe);
+		if (qhp && flush_qhp != qhp)
+			spin_unlock(&qhp->lock);
 	}
 }
 
diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c
index feeb8ee6f4a2..44161ca4d2a8 100644
--- a/drivers/infiniband/hw/cxgb4/device.c
+++ b/drivers/infiniband/hw/cxgb4/device.c
@@ -875,6 +875,11 @@ static int c4iw_rdev_open(struct c4iw_rdev *rdev)
 
 	rdev->status_page->db_off = 0;
 
+	init_completion(&rdev->rqt_compl);
+	init_completion(&rdev->pbl_compl);
+	kref_init(&rdev->rqt_kref);
+	kref_init(&rdev->pbl_kref);
+
 	return 0;
 err_free_status_page_and_wr_log:
 	if (c4iw_wr_log && rdev->wr_log)
@@ -893,13 +898,15 @@ destroy_resource:
 
 static void c4iw_rdev_close(struct c4iw_rdev *rdev)
 {
-	destroy_workqueue(rdev->free_workq);
 	kfree(rdev->wr_log);
 	c4iw_release_dev_ucontext(rdev, &rdev->uctx);
 	free_page((unsigned long)rdev->status_page);
 	c4iw_pblpool_destroy(rdev);
 	c4iw_rqtpool_destroy(rdev);
+	wait_for_completion(&rdev->pbl_compl);
+	wait_for_completion(&rdev->rqt_compl);
 	c4iw_ocqp_pool_destroy(rdev);
+	destroy_workqueue(rdev->free_workq);
 	c4iw_destroy_resource(&rdev->resource);
 }
 
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index 5eec8772468c..870649ff049c 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -186,6 +186,10 @@ struct c4iw_rdev {
 	struct wr_log_entry *wr_log;
 	int wr_log_size;
 	struct workqueue_struct *free_workq;
+	struct completion rqt_compl;
+	struct completion pbl_compl;
+	struct kref rqt_kref;
+	struct kref pbl_kref;
 };
 
 static inline int c4iw_fatal_error(struct c4iw_rdev *rdev)
@@ -1050,7 +1054,7 @@ u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size);
 void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size);
 u32 c4iw_ocqp_pool_alloc(struct c4iw_rdev *rdev, int size);
 void c4iw_ocqp_pool_free(struct c4iw_rdev *rdev, u32 addr, int size);
-void c4iw_flush_hw_cq(struct c4iw_cq *chp);
+void c4iw_flush_hw_cq(struct c4iw_cq *chp, struct c4iw_qp *flush_qhp);
 void c4iw_count_rcqes(struct t4_cq *cq, struct t4_wq *wq, int *count);
 int c4iw_ep_disconnect(struct c4iw_ep *ep, int abrupt, gfp_t gfp);
 int c4iw_flush_rq(struct t4_wq *wq, struct t4_cq *cq, int count);
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index 2dc94997ea11..4106eed1b8fb 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -1342,12 +1342,12 @@ static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp,
 	qhp->wq.flushed = 1;
 	t4_set_wq_in_error(&qhp->wq);
 
-	c4iw_flush_hw_cq(rchp);
+	c4iw_flush_hw_cq(rchp, qhp);
 	c4iw_count_rcqes(&rchp->cq, &qhp->wq, &count);
 	rq_flushed = c4iw_flush_rq(&qhp->wq, &rchp->cq, count);
 
 	if (schp != rchp)
-		c4iw_flush_hw_cq(schp);
+		c4iw_flush_hw_cq(schp, qhp);
 	sq_flushed = c4iw_flush_sq(qhp);
 
 	spin_unlock(&qhp->lock);
diff --git a/drivers/infiniband/hw/cxgb4/resource.c b/drivers/infiniband/hw/cxgb4/resource.c
index 3cf25997ed2b..0ef25ae05e6f 100644
--- a/drivers/infiniband/hw/cxgb4/resource.c
+++ b/drivers/infiniband/hw/cxgb4/resource.c
@@ -260,12 +260,22 @@ u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size)
 		rdev->stats.pbl.cur += roundup(size, 1 << MIN_PBL_SHIFT);
 		if (rdev->stats.pbl.cur > rdev->stats.pbl.max)
 			rdev->stats.pbl.max = rdev->stats.pbl.cur;
+		kref_get(&rdev->pbl_kref);
 	} else
 		rdev->stats.pbl.fail++;
 	mutex_unlock(&rdev->stats.lock);
 	return (u32)addr;
 }
 
+static void destroy_pblpool(struct kref *kref)
+{
+	struct c4iw_rdev *rdev;
+
+	rdev = container_of(kref, struct c4iw_rdev, pbl_kref);
+	gen_pool_destroy(rdev->pbl_pool);
+	complete(&rdev->pbl_compl);
+}
+
 void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 {
 	pr_debug("addr 0x%x size %d\n", addr, size);
@@ -273,6 +283,7 @@ void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 	rdev->stats.pbl.cur -= roundup(size, 1 << MIN_PBL_SHIFT);
 	mutex_unlock(&rdev->stats.lock);
 	gen_pool_free(rdev->pbl_pool, (unsigned long)addr, size);
+	kref_put(&rdev->pbl_kref, destroy_pblpool);
 }
 
 int c4iw_pblpool_create(struct c4iw_rdev *rdev)
@@ -310,7 +321,7 @@ int c4iw_pblpool_create(struct c4iw_rdev *rdev)
 
 void c4iw_pblpool_destroy(struct c4iw_rdev *rdev)
 {
-	gen_pool_destroy(rdev->pbl_pool);
+	kref_put(&rdev->pbl_kref, destroy_pblpool);
 }
 
 /*
@@ -331,12 +342,22 @@ u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size)
 		rdev->stats.rqt.cur += roundup(size << 6, 1 << MIN_RQT_SHIFT);
 		if (rdev->stats.rqt.cur > rdev->stats.rqt.max)
 			rdev->stats.rqt.max = rdev->stats.rqt.cur;
+		kref_get(&rdev->rqt_kref);
 	} else
 		rdev->stats.rqt.fail++;
 	mutex_unlock(&rdev->stats.lock);
 	return (u32)addr;
 }
 
+static void destroy_rqtpool(struct kref *kref)
+{
+	struct c4iw_rdev *rdev;
+
+	rdev = container_of(kref, struct c4iw_rdev, rqt_kref);
+	gen_pool_destroy(rdev->rqt_pool);
+	complete(&rdev->rqt_compl);
+}
+
 void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 {
 	pr_debug("addr 0x%x size %d\n", addr, size << 6);
@@ -344,6 +365,7 @@ void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 	rdev->stats.rqt.cur -= roundup(size << 6, 1 << MIN_RQT_SHIFT);
 	mutex_unlock(&rdev->stats.lock);
 	gen_pool_free(rdev->rqt_pool, (unsigned long)addr, size << 6);
+	kref_put(&rdev->rqt_kref, destroy_rqtpool);
 }
 
 int c4iw_rqtpool_create(struct c4iw_rdev *rdev)
@@ -380,7 +402,7 @@ int c4iw_rqtpool_create(struct c4iw_rdev *rdev)
 
 void c4iw_rqtpool_destroy(struct c4iw_rdev *rdev)
 {
-	gen_pool_destroy(rdev->rqt_pool);
+	kref_put(&rdev->rqt_kref, destroy_rqtpool);
 }
 
 /*