All of lore.kernel.org
 help / color / mirror / Atom feed
From: Tom Tucker <tom@opengridcomputing.com>
To: bfields@fieldses.org
Cc: linux-nfs@vger.kernel.org, Tom Tucker <tom@opengridcomputing.com>
Subject: [PATCH 04/11] svcrdma: Move the DMA unmap logic to the CQ handler
Date: Wed,  2 Jul 2008 21:27:36 -0500	[thread overview]
Message-ID: <1215052063-7330-5-git-send-email-tom@opengridcomputing.com> (raw)
In-Reply-To: <1215052063-7330-4-git-send-email-tom@opengridcomputing.com>

Separate DMA unmap from context destruction and perform DMA unmapping
in the SQ/RQ CQ reap functions. This is necessary to support software
based RDMA implementations that actually copy the data in their
ib_dma_unmap callback functions and architectures that don't have
cache coherent I/O busses.

Signed-off-by: Tom Tucker <tom@opengridcomputing.com>

---
 net/sunrpc/xprtrdma/svc_rdma_transport.c |   20 ++++++++++++++------
 1 files changed, 14 insertions(+), 6 deletions(-)

diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index fc86338..7e8ee66 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -150,6 +150,18 @@ struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *xprt)
 	return ctxt;
 }
 
+static void svc_rdma_unmap_dma(struct svc_rdma_op_ctxt *ctxt)
+{
+	struct svcxprt_rdma *xprt = ctxt->xprt;
+	int i;
+	for (i = 0; i < ctxt->count && ctxt->sge[i].length; i++) {
+		ib_dma_unmap_single(xprt->sc_cm_id->device,
+				    ctxt->sge[i].addr,
+				    ctxt->sge[i].length,
+				    ctxt->direction);
+	}
+}
+
 void svc_rdma_put_context(struct svc_rdma_op_ctxt *ctxt, int free_pages)
 {
 	struct svcxprt_rdma *xprt;
@@ -161,12 +173,6 @@ void svc_rdma_put_context(struct svc_rdma_op_ctxt *ctxt, int free_pages)
 		for (i = 0; i < ctxt->count; i++)
 			put_page(ctxt->pages[i]);
 
-	for (i = 0; i < ctxt->count; i++)
-		ib_dma_unmap_single(xprt->sc_cm_id->device,
-				    ctxt->sge[i].addr,
-				    ctxt->sge[i].length,
-				    ctxt->direction);
-
 	spin_lock_bh(&xprt->sc_ctxt_lock);
 	list_add(&ctxt->free_list, &xprt->sc_ctxt_free);
 	spin_unlock_bh(&xprt->sc_ctxt_lock);
@@ -328,6 +334,7 @@ static void rq_cq_reap(struct svcxprt_rdma *xprt)
 		ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id;
 		ctxt->wc_status = wc.status;
 		ctxt->byte_len = wc.byte_len;
+		svc_rdma_unmap_dma(ctxt);
 		if (wc.status != IB_WC_SUCCESS) {
 			/* Close the transport */
 			dprintk("svcrdma: transport closing putting ctxt %p\n", ctxt);
@@ -377,6 +384,7 @@ static void sq_cq_reap(struct svcxprt_rdma *xprt)
 		ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id;
 		xprt = ctxt->xprt;
 
+		svc_rdma_unmap_dma(ctxt);
 		if (wc.status != IB_WC_SUCCESS)
 			/* Close the transport */
 			set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);

  reply	other threads:[~2008-07-03  2:27 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-07-03  2:27 [PATCH 00/11] svcrdma: WR context management bug fixes and cleanup Tom Tucker
2008-07-03  2:27 ` [PATCH 01/11] svcrdma: Add a type for keeping NFS RPC mapping Tom Tucker
2008-07-03  2:27   ` [PATCH 02/11] svcrdma: Use RPC reply map for RDMA_WRITE processing Tom Tucker
2008-07-03  2:27     ` [PATCH 03/11] svcrdma: Use reply and chunk map for RDMA_READ processing Tom Tucker
2008-07-03  2:27       ` Tom Tucker [this message]
2008-07-03  2:27         ` [PATCH 05/11] svcrdma: Add dma map count and WARN_ON Tom Tucker
2008-07-03  2:27           ` [PATCH 06/11] svcrdma: Remove unneeded spin locks from __svc_rdma_free Tom Tucker
2008-07-03  2:27             ` [PATCH 07/11] svcrdma: Remove unused wait q from svcrdma_xprt structure Tom Tucker
2008-07-03  2:27               ` [PATCH 08/11] svcrdma: Limit ORD based on client's advertised IRD Tom Tucker
2008-07-03  2:27                 ` [PATCH 09/11] svcrdma: Add flush_scheduled_work to module exit function Tom Tucker
2008-07-03  2:27                   ` [PATCH 10/11] svcrdma: Create a kmem cache for the WR contexts Tom Tucker
2008-07-03  2:27                     ` [PATCH 11/11] svcrdma: Change WR context get/put to use the kmem cache Tom Tucker
2008-07-03 20:27 ` [PATCH 00/11] svcrdma: WR context management bug fixes and cleanup J. Bruce Fields

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1215052063-7330-5-git-send-email-tom@opengridcomputing.com \
    --to=tom@opengridcomputing.com \
    --cc=bfields@fieldses.org \
    --cc=linux-nfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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.