* [PATCH] RDMA/cxgb3: Don't allocate the sw queue for user mode CQs.
@ 2010-01-27 17:03 Steve Wise
[not found] ` <20100127170353.19882.40656.stgit-T4OLL4TyM9aNDNWfRnPdfg@public.gmane.org>
0 siblings, 1 reply; 2+ messages in thread
From: Steve Wise @ 2010-01-27 17:03 UTC (permalink / raw)
To: rdreier-FYB4Gu1CFyUAvxtiuMwx3w; +Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA
Only kernel mode CQs need the sw queue memory allocated. The sw queue
for user mode CQs is allocated in user mode by libcxgb3.
Signed-off-by: Steve Wise <swise-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW@public.gmane.org>
---
drivers/infiniband/hw/cxgb3/cxio_hal.c | 20 +++++++++++---------
drivers/infiniband/hw/cxgb3/cxio_hal.h | 2 +-
drivers/infiniband/hw/cxgb3/iwch_provider.c | 2 +-
3 files changed, 13 insertions(+), 11 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
index 8837d56..396da7b 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
@@ -154,7 +154,7 @@ static int cxio_hal_clear_qp_ctx(struct cxio_rdev *rdev_p, u32 qpid)
return iwch_cxgb3_ofld_send(rdev_p->t3cdev_p, skb);
}
-int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
+int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq, int kernel)
{
struct rdma_cq_setup setup;
int size = (1UL << (cq->size_log2)) * sizeof(struct t3_cqe);
@@ -162,15 +162,16 @@ int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
cq->cqid = cxio_hal_get_cqid(rdev_p->rscp);
if (!cq->cqid)
return -ENOMEM;
- cq->sw_queue = kzalloc(size, GFP_KERNEL);
- if (!cq->sw_queue)
- return -ENOMEM;
- cq->queue = dma_alloc_coherent(&(rdev_p->rnic_info.pdev->dev),
- (1UL << (cq->size_log2)) *
- sizeof(struct t3_cqe),
+ if (kernel) {
+ cq->sw_queue = kzalloc(size, GFP_KERNEL);
+ if (!cq->sw_queue)
+ return -ENOMEM;
+ }
+ cq->queue = dma_alloc_coherent(&(rdev_p->rnic_info.pdev->dev), size,
&(cq->dma_addr), GFP_KERNEL);
if (!cq->queue) {
- kfree(cq->sw_queue);
+ if (kernel)
+ kfree(cq->sw_queue);
return -ENOMEM;
}
pci_unmap_addr_set(cq, mapping, cq->dma_addr);
@@ -320,7 +321,8 @@ int cxio_destroy_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
{
int err;
err = cxio_hal_clear_cq_ctx(rdev_p, cq->cqid);
- kfree(cq->sw_queue);
+ if (cq->sw_queue)
+ kfree(cq->sw_queue);
dma_free_coherent(&(rdev_p->rnic_info.pdev->dev),
(1UL << (cq->size_log2))
* sizeof(struct t3_cqe), cq->queue,
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.h b/drivers/infiniband/hw/cxgb3/cxio_hal.h
index f3d440c..baa2867 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.h
@@ -157,7 +157,7 @@ int cxio_rdev_open(struct cxio_rdev *rdev);
void cxio_rdev_close(struct cxio_rdev *rdev);
int cxio_hal_cq_op(struct cxio_rdev *rdev, struct t3_cq *cq,
enum t3_cq_opcode op, u32 credit);
-int cxio_create_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
+int cxio_create_cq(struct cxio_rdev *rdev, struct t3_cq *cq, int kernel);
int cxio_destroy_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
int cxio_resize_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
void cxio_release_ucontext(struct cxio_rdev *rdev, struct cxio_ucontext *uctx);
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index ed71755..47b35c6 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -187,7 +187,7 @@ static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int ve
entries = roundup_pow_of_two(entries);
chp->cq.size_log2 = ilog2(entries);
- if (cxio_create_cq(&rhp->rdev, &chp->cq)) {
+ if (cxio_create_cq(&rhp->rdev, &chp->cq, !ucontext)) {
kfree(chp);
return ERR_PTR(-ENOMEM);
}
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH] RDMA/cxgb3: Don't allocate the sw queue for user mode CQs.
[not found] ` <20100127170353.19882.40656.stgit-T4OLL4TyM9aNDNWfRnPdfg@public.gmane.org>
@ 2010-01-27 19:39 ` Roland Dreier
0 siblings, 0 replies; 2+ messages in thread
From: Roland Dreier @ 2010-01-27 19:39 UTC (permalink / raw)
To: Steve Wise; +Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA
> - kfree(cq->sw_queue);
> + if (cq->sw_queue)
> + kfree(cq->sw_queue);
Looks like cq->sw_queue will be set to NULL when it's not allocated (ie
the userspace case). So that's OK.
But kfree(NULL) is fine. So you can drop this hunk I think.
- R.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2010-01-27 19:39 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-01-27 17:03 [PATCH] RDMA/cxgb3: Don't allocate the sw queue for user mode CQs Steve Wise
[not found] ` <20100127170353.19882.40656.stgit-T4OLL4TyM9aNDNWfRnPdfg@public.gmane.org>
2010-01-27 19:39 ` Roland Dreier
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.