[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20131202191257.897405691@linuxfoundation.org>
Date: Mon, 2 Dec 2013 11:14:44 -0800
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
stable@...r.kernel.org, Vu Pham <vu@...lanox.com>,
Nicholas Bellinger <nab@...ux-iscsi.org>
Subject: [PATCH 3.12 093/212] iser-target: Avoid using FRMR for single dma entry requests
3.12-stable review patch. If anyone has any objections, please let me know.
------------------
From: Vu Pham <vu@...lanox.com>
commit f01b9f73392b48c6cda7c2c66594c73137c776da upstream.
This patch changes isert_reg_rdma_frwr() to not use FRMR for single
dma entry requests from small I/Os, in order to avoid the associated
memory registration overhead.
Using DMA MR is sufficient here for the single dma entry requests,
and addresses a >= v3.12 performance regression.
Signed-off-by: Vu Pham <vu@...lanox.com>
Signed-off-by: Nicholas Bellinger <nab@...ux-iscsi.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
drivers/infiniband/ulp/isert/ib_isert.c | 30 +++++++++++++++++++-----------
1 file changed, 19 insertions(+), 11 deletions(-)
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -2259,18 +2259,26 @@ isert_reg_rdma_frwr(struct iscsi_conn *c
data_len = min(data_left, rdma_write_max);
wr->cur_rdma_length = data_len;
- spin_lock_irqsave(&isert_conn->conn_lock, flags);
- fr_desc = list_first_entry(&isert_conn->conn_frwr_pool,
- struct fast_reg_descriptor, list);
- list_del(&fr_desc->list);
- spin_unlock_irqrestore(&isert_conn->conn_lock, flags);
- wr->fr_desc = fr_desc;
+ /* if there is a single dma entry, dma mr is sufficient */
+ if (count == 1) {
+ ib_sge->addr = ib_sg_dma_address(ib_dev, &sg_start[0]);
+ ib_sge->length = ib_sg_dma_len(ib_dev, &sg_start[0]);
+ ib_sge->lkey = isert_conn->conn_mr->lkey;
+ wr->fr_desc = NULL;
+ } else {
+ spin_lock_irqsave(&isert_conn->conn_lock, flags);
+ fr_desc = list_first_entry(&isert_conn->conn_frwr_pool,
+ struct fast_reg_descriptor, list);
+ list_del(&fr_desc->list);
+ spin_unlock_irqrestore(&isert_conn->conn_lock, flags);
+ wr->fr_desc = fr_desc;
- ret = isert_fast_reg_mr(fr_desc, isert_cmd, isert_conn,
- ib_sge, offset, data_len);
- if (ret) {
- list_add_tail(&fr_desc->list, &isert_conn->conn_frwr_pool);
- goto unmap_sg;
+ ret = isert_fast_reg_mr(fr_desc, isert_cmd, isert_conn,
+ ib_sge, offset, data_len);
+ if (ret) {
+ list_add_tail(&fr_desc->list, &isert_conn->conn_frwr_pool);
+ goto unmap_sg;
+ }
}
return 0;
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists