From: Christoph Hellwig <hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
To: Sagi Grimberg <sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: [PATCH RFC 2/3] svcrdma: Use device rdma_read_access_flags
Date: Tue, 10 Nov 2015 04:04:32 -0800 [thread overview]
Message-ID: <20151110120432.GA8230@infradead.org> (raw)
In-Reply-To: <5641D920.5000409-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
On Tue, Nov 10, 2015 at 01:46:40PM +0200, Sagi Grimberg wrote:
>
>
> On 10/11/2015 13:41, Christoph Hellwig wrote:
> >Oh, and while we're at it. Can someone explain why we're even
> >using rdma_read_chunk_frmr for IB? It seems to work around the
> >fact tat iWarp only allow a single RDMA READ SGE, but it's used
> >whenever the device has IB_DEVICE_MEM_MGT_EXTENSIONS, which seems
> >wrong.
>
> I think Steve can answer it better than I can. I think that it is
> just to have a single code path for both IB and iWARP. I agree that
> the condition seems wrong and for small transfers rdma_read_chunk_frmr
> is really a performance loss.
Well, the code path already exists, but only is used fi
IB_DEVICE_MEM_MGT_EXTENSIONS isn't set. Below is an untested patch
that demonstrates how I think svcrdma should setup the reads. Note
that this also allows to entirely remove it's allphys MR.
Note that as a followon this would also allow to remove the
non-READ_W_INV code path from rdma_read_chunk_frmr as a future
step.
diff --git a/include/linux/sunrpc/svc_rdma.h b/include/linux/sunrpc/svc_rdma.h
index f869807..35fa638 100644
--- a/include/linux/sunrpc/svc_rdma.h
+++ b/include/linux/sunrpc/svc_rdma.h
@@ -147,7 +147,6 @@ struct svcxprt_rdma {
struct ib_qp *sc_qp;
struct ib_cq *sc_rq_cq;
struct ib_cq *sc_sq_cq;
- struct ib_mr *sc_phys_mr; /* MR for server memory */
int (*sc_reader)(struct svcxprt_rdma *,
struct svc_rqst *,
struct svc_rdma_op_ctxt *,
diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
index ff4f01e..a410cb6 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
@@ -160,7 +160,7 @@ int rdma_read_chunk_lcl(struct svcxprt_rdma *xprt,
goto err;
atomic_inc(&xprt->sc_dma_used);
- /* The lkey here is either a local dma lkey or a dma_mr lkey */
+ /* The lkey here is the local dma lkey */
ctxt->sge[pno].lkey = xprt->sc_dma_lkey;
ctxt->sge[pno].length = len;
ctxt->count++;
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index 9f3eb89..2780da4 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -887,8 +887,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
struct ib_cq_init_attr cq_attr = {};
struct ib_qp_init_attr qp_attr;
struct ib_device *dev;
- int uninitialized_var(dma_mr_acc);
- int need_dma_mr = 0;
int ret = 0;
int i;
@@ -986,68 +984,41 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
}
newxprt->sc_qp = newxprt->sc_cm_id->qp;
- /*
- * Use the most secure set of MR resources based on the
- * transport type and available memory management features in
- * the device. Here's the table implemented below:
- *
- * Fast Global DMA Remote WR
- * Reg LKEY MR Access
- * Sup'd Sup'd Needed Needed
- *
- * IWARP N N Y Y
- * N Y Y Y
- * Y N Y N
- * Y Y N -
- *
- * IB N N Y N
- * N Y N -
- * Y N Y N
- * Y Y N -
- *
- * NB: iWARP requires remote write access for the data sink
- * of an RDMA_READ. IB does not.
- */
- newxprt->sc_reader = rdma_read_chunk_lcl;
- if (dev->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) {
- newxprt->sc_frmr_pg_list_len =
- dev->max_fast_reg_page_list_len;
- newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_FAST_REG;
- newxprt->sc_reader = rdma_read_chunk_frmr;
- }
+ if (rdma_protocol_iwarp(dev, newxprt->sc_cm_id->port_num)) {
+ /*
+ * iWARP requires remote write access for the data sink, and
+ * only supports a single SGE for RDMA_READ requests, so we'll
+ * have to use a memory registration for each RDMA_READ.
+ */
+ if (!(dev->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS)) {
+ /*
+ * We're an iWarp device but don't support FRs.
+ * Tought luck, better exit early as we have little
+ * chance of doing something useful.
+ */
+ goto errout;
+ }
- /*
- * Determine if a DMA MR is required and if so, what privs are required
- */
- if (!rdma_protocol_iwarp(dev, newxprt->sc_cm_id->port_num) &&
- !rdma_ib_or_roce(dev, newxprt->sc_cm_id->port_num))
+ newxprt->sc_frmr_pg_list_len = dev->max_fast_reg_page_list_len;
+ newxprt->sc_dev_caps =
+ SVCRDMA_DEVCAP_FAST_REG |
+ SVCRDMA_DEVCAP_READ_W_INV;
+ newxprt->sc_reader = rdma_read_chunk_frmr;
+ } else if (rdma_ib_or_roce(dev, newxprt->sc_cm_id->port_num)) {
+ /*
+ * For IB or RoCE life is easy, no unsafe write access is
+ * required and multiple SGEs are supported, so we don't need
+ * to use MRs.
+ */
+ newxprt->sc_reader = rdma_read_chunk_lcl;
+ } else {
+ /*
+ * Neither iWarp nor IB-ish, we're out of luck.
+ */
goto errout;
-
- if (!(newxprt->sc_dev_caps & SVCRDMA_DEVCAP_FAST_REG) ||
- !(dev->device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY)) {
- need_dma_mr = 1;
- dma_mr_acc = IB_ACCESS_LOCAL_WRITE;
- if (rdma_protocol_iwarp(dev, newxprt->sc_cm_id->port_num) &&
- !(newxprt->sc_dev_caps & SVCRDMA_DEVCAP_FAST_REG))
- dma_mr_acc |= IB_ACCESS_REMOTE_WRITE;
}
- if (rdma_protocol_iwarp(dev, newxprt->sc_cm_id->port_num))
- newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_READ_W_INV;
-
- /* Create the DMA MR if needed, otherwise, use the DMA LKEY */
- if (need_dma_mr) {
- /* Register all of physical memory */
- newxprt->sc_phys_mr =
- ib_get_dma_mr(newxprt->sc_pd, dma_mr_acc);
- if (IS_ERR(newxprt->sc_phys_mr)) {
- dprintk("svcrdma: Failed to create DMA MR ret=%d\n",
- ret);
- goto errout;
- }
- newxprt->sc_dma_lkey = newxprt->sc_phys_mr->lkey;
- } else
- newxprt->sc_dma_lkey = dev->local_dma_lkey;
+ newxprt->sc_dma_lkey = dev->local_dma_lkey;
/* Post receive buffers */
for (i = 0; i < newxprt->sc_max_requests; i++) {
@@ -1203,9 +1174,6 @@ static void __svc_rdma_free(struct work_struct *work)
if (rdma->sc_rq_cq && !IS_ERR(rdma->sc_rq_cq))
ib_destroy_cq(rdma->sc_rq_cq);
- if (rdma->sc_phys_mr && !IS_ERR(rdma->sc_phys_mr))
- ib_dereg_mr(rdma->sc_phys_mr);
-
if (rdma->sc_pd && !IS_ERR(rdma->sc_pd))
ib_dealloc_pd(rdma->sc_pd);
--
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
next prev parent reply other threads:[~2015-11-10 12:04 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-10 10:44 [PATCH RFC 0/3] Introduce device attribute rdma_read_access_flags Sagi Grimberg
[not found] ` <1447152255-28231-1-git-send-email-sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 10:44 ` [PATCH RFC 1/3] IB/core: Expose a device attribute for rdma_read access flags Sagi Grimberg
[not found] ` <1447152255-28231-2-git-send-email-sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 11:25 ` Christoph Hellwig
[not found] ` <20151110112537.GA8991-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 11:36 ` Sagi Grimberg
2015-11-10 11:51 ` Yann Droneaud
[not found] ` <1447156270.7089.3.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org>
2015-11-10 12:28 ` Sagi Grimberg
[not found] ` <5641E2D1.8070209-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-10 12:39 ` Sagi Grimberg
2015-11-10 12:36 ` Tom Talpey
[not found] ` <5641E4C9.7000206-CLs1Zie5N5HQT0dZR+AlfA@public.gmane.org>
2015-11-10 12:42 ` Sagi Grimberg
[not found] ` <5641E644.7080101-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 13:06 ` Christoph Hellwig
[not found] ` <20151110130648.GA10682-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 13:41 ` Christoph Hellwig
[not found] ` <20151110134147.GA12814-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 18:36 ` Jason Gunthorpe
[not found] ` <20151110183627.GJ12667-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-11-11 8:03 ` Christoph Hellwig
2015-11-11 9:07 ` Sagi Grimberg
[not found] ` <56430542.5090008-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-11 10:25 ` Christoph Hellwig
2015-11-10 18:06 ` Jason Gunthorpe
2015-11-10 18:01 ` Jason Gunthorpe
[not found] ` <20151110180156.GE12667-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-11-11 8:08 ` Christoph Hellwig
[not found] ` <20151111080837.GA14662-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-11 9:09 ` Sagi Grimberg
[not found] ` <564305AF.3020903-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-11 10:26 ` Christoph Hellwig
2015-11-10 10:44 ` [PATCH RFC 2/3] svcrdma: Use device rdma_read_access_flags Sagi Grimberg
[not found] ` <1447152255-28231-3-git-send-email-sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 11:38 ` Christoph Hellwig
[not found] ` <20151110113839.GA32523-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 11:42 ` Sagi Grimberg
2015-11-10 17:03 ` Chuck Lever
2015-11-10 11:41 ` Christoph Hellwig
[not found] ` <20151110114145.GA2810-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 11:46 ` Sagi Grimberg
[not found] ` <5641D920.5000409-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 12:04 ` Christoph Hellwig [this message]
[not found] ` <20151110120432.GA8230-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 14:34 ` Chuck Lever
[not found] ` <F60E8BA6-D728-44E4-9331-0C23AB96E634-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-10 15:16 ` Tom Talpey
2015-11-10 18:25 ` Jason Gunthorpe
[not found] ` <20151110182546.GI12667-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-11-10 21:00 ` Tom Talpey
[not found] ` <56425AFB.30202-CLs1Zie5N5HQT0dZR+AlfA@public.gmane.org>
2015-11-10 21:17 ` Jason Gunthorpe
[not found] ` <20151110211716.GA21631-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-11-10 21:30 ` Tom Talpey
[not found] ` <564261EF.4000008-CLs1Zie5N5HQT0dZR+AlfA@public.gmane.org>
2015-11-10 23:01 ` Chuck Lever
[not found] ` <54CF0111-E9C4-4196-BF92-7E134A6A329A-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-10 23:55 ` Jason Gunthorpe
[not found] ` <20151110235551.GA12795-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-11-11 8:07 ` Christoph Hellwig
2015-11-11 9:28 ` Sagi Grimberg
[not found] ` <56430A20.6040600-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-11 16:19 ` Chuck Lever
[not found] ` <D4228973-6DFA-4181-9848-41BFE8E75200-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-11 16:29 ` Sagi Grimberg
[not found] ` <56436CDB.6090305-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-11 20:50 ` Chuck Lever
2015-11-11 9:25 ` Sagi Grimberg
[not found] ` <56430972.8080703-LDSdmyG8hGV8YrgS2mwiifqBs+8SCbDb@public.gmane.org>
2015-11-11 17:39 ` Jason Gunthorpe
2015-11-11 8:02 ` Christoph Hellwig
[not found] ` <20151111080225.GA31508-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-11 17:23 ` Jason Gunthorpe
2015-11-10 16:12 ` Steve Wise
2015-11-10 10:44 ` [PATCH RFC 3/3] RDS_IW: " Sagi Grimberg
[not found] ` <1447152255-28231-4-git-send-email-sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
2015-11-10 11:26 ` Christoph Hellwig
[not found] ` <20151110112638.GB8991-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 11:35 ` Sagi Grimberg
2015-11-10 11:31 ` [PATCH RFC 0/3] Introduce device attribute rdma_read_access_flags Christoph Hellwig
[not found] ` <20151110113141.GA23093-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
2015-11-10 11:40 ` Sagi Grimberg
2015-11-10 16:18 ` Steve Wise
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=20151110120432.GA8230@infradead.org \
--to=hch-wegcikhe2lqwvfeawa7xhq@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).