All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 14:44 Chuck Lever
  2015-11-12 14:47   ` Chuck Lever
  2015-11-12 19:13 ` J. Bruce Fields
  0 siblings, 2 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-12 14:44 UTC (permalink / raw)
  To: bfields; +Cc: linux-nfs, linux-rmda

Minor optimization: when dealing with write chunk XDR roundup, do
not post a Write WR for the zero bytes in the pad. Simply update
the write segment in the RPC-over-RDMA header to reflect the extra
pad bytes.

The Reply chunk is also a write chunk, but the server does not use
send_write_chunks() to send the Reply chunk. That's OK in this case:
the server Upper Layer typically marshals the Reply chunk contents
in a single contiguous buffer, without a separate tail for the XDR
pad.

The comments and the variable naming refer to "chunks" but what is
really meant is "segments." The existing code sends only one
xdr_write_chunk per RPC reply.

The fix assumes this as well. When the XDR pad in the first write
chunk is reached, the assumption is the Write list is complete and
send_write_chunks() returns.

That will remain a valid assumption until the server Upper Layer can
support multiple bulk payload results per RPC.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
 net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 1dfae83..6c48901 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
 						arg_ch->rs_handle,
 						arg_ch->rs_offset,
 						write_len);
+
+		/* Do not send XDR pad bytes */
+		if (chunk_no && write_len < 4) {
+			chunk_no++;
+			break;
+		}
+
 		chunk_off = 0;
 		while (write_len) {
 			ret = send_write(xprt, rqstp,


^ permalink raw reply related	[flat|nested] 18+ messages in thread

* Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-12 14:44 [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk Chuck Lever
@ 2015-11-12 14:47   ` Chuck Lever
  2015-11-12 19:13 ` J. Bruce Fields
  1 sibling, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-12 14:47 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux RDMA Mailing List, Linux NFS Mailing List

I wish git send-mail had an address book. I’m really tired
of misspelling the to/cc addresses on patches.

Resending so there is one thread for all replies. Apologies
for the noise.


> Begin forwarded message:
> 
> From: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> Subject: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
> Date: November 12, 2015 at 9:44:33 AM EST
> To: bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org
> Cc: linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-rmda-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> 
> Minor optimization: when dealing with write chunk XDR roundup, do
> not post a Write WR for the zero bytes in the pad. Simply update
> the write segment in the RPC-over-RDMA header to reflect the extra
> pad bytes.
> 
> The Reply chunk is also a write chunk, but the server does not use
> send_write_chunks() to send the Reply chunk. That's OK in this case:
> the server Upper Layer typically marshals the Reply chunk contents
> in a single contiguous buffer, without a separate tail for the XDR
> pad.
> 
> The comments and the variable naming refer to "chunks" but what is
> really meant is "segments." The existing code sends only one
> xdr_write_chunk per RPC reply.
> 
> The fix assumes this as well. When the XDR pad in the first write
> chunk is reached, the assumption is the Write list is complete and
> send_write_chunks() returns.
> 
> That will remain a valid assumption until the server Upper Layer can
> support multiple bulk payload results per RPC.
> 
> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> ---
> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> 1 file changed, 7 insertions(+)
> 
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> index 1dfae83..6c48901 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> 						arg_ch->rs_handle,
> 						arg_ch->rs_offset,
> 						write_len);
> +
> +		/* Do not send XDR pad bytes */
> +		if (chunk_no && write_len < 4) {
> +			chunk_no++;
> +			break;
> +		}
> +
> 		chunk_off = 0;
> 		while (write_len) {
> 			ret = send_write(xprt, rqstp,
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
Chuck Lever




--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" 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] 18+ messages in thread

* Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 14:47   ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-12 14:47 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux RDMA Mailing List, Linux NFS Mailing List

I wish git send-mail had an address book. I’m really tired
of misspelling the to/cc addresses on patches.

Resending so there is one thread for all replies. Apologies
for the noise.


> Begin forwarded message:
> 
> From: Chuck Lever <chuck.lever@oracle.com>
> Subject: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
> Date: November 12, 2015 at 9:44:33 AM EST
> To: bfields@fieldses.org
> Cc: linux-nfs@vger.kernel.org, linux-rmda@vger.kernel.org
> 
> Minor optimization: when dealing with write chunk XDR roundup, do
> not post a Write WR for the zero bytes in the pad. Simply update
> the write segment in the RPC-over-RDMA header to reflect the extra
> pad bytes.
> 
> The Reply chunk is also a write chunk, but the server does not use
> send_write_chunks() to send the Reply chunk. That's OK in this case:
> the server Upper Layer typically marshals the Reply chunk contents
> in a single contiguous buffer, without a separate tail for the XDR
> pad.
> 
> The comments and the variable naming refer to "chunks" but what is
> really meant is "segments." The existing code sends only one
> xdr_write_chunk per RPC reply.
> 
> The fix assumes this as well. When the XDR pad in the first write
> chunk is reached, the assumption is the Write list is complete and
> send_write_chunks() returns.
> 
> That will remain a valid assumption until the server Upper Layer can
> support multiple bulk payload results per RPC.
> 
> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> ---
> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> 1 file changed, 7 insertions(+)
> 
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> index 1dfae83..6c48901 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> 						arg_ch->rs_handle,
> 						arg_ch->rs_offset,
> 						write_len);
> +
> +		/* Do not send XDR pad bytes */
> +		if (chunk_no && write_len < 4) {
> +			chunk_no++;
> +			break;
> +		}
> +
> 		chunk_off = 0;
> 		while (write_len) {
> 			ret = send_write(xprt, rqstp,
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
Chuck Lever





^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-12 14:47   ` Chuck Lever
@ 2015-11-12 15:17       ` J. Bruce Fields
  -1 siblings, 0 replies; 18+ messages in thread
From: J. Bruce Fields @ 2015-11-12 15:17 UTC (permalink / raw)
  To: Chuck Lever; +Cc: Linux RDMA Mailing List, Linux NFS Mailing List

On Thu, Nov 12, 2015 at 09:47:13AM -0500, Chuck Lever wrote:
> I wish git send-mail had an address book. I’m really tired
> of misspelling the to/cc addresses on patches.

Hah.  Yeah, my hack is a text file of addresses that I cut-and-paste
from.

--b.

> 
> Resending so there is one thread for all replies. Apologies
> for the noise.
> 
> 
> > Begin forwarded message:
> > 
> > From: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> > Subject: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
> > Date: November 12, 2015 at 9:44:33 AM EST
> > To: bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org
> > Cc: linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-rmda-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> > 
> > Minor optimization: when dealing with write chunk XDR roundup, do
> > not post a Write WR for the zero bytes in the pad. Simply update
> > the write segment in the RPC-over-RDMA header to reflect the extra
> > pad bytes.
> > 
> > The Reply chunk is also a write chunk, but the server does not use
> > send_write_chunks() to send the Reply chunk. That's OK in this case:
> > the server Upper Layer typically marshals the Reply chunk contents
> > in a single contiguous buffer, without a separate tail for the XDR
> > pad.
> > 
> > The comments and the variable naming refer to "chunks" but what is
> > really meant is "segments." The existing code sends only one
> > xdr_write_chunk per RPC reply.
> > 
> > The fix assumes this as well. When the XDR pad in the first write
> > chunk is reached, the assumption is the Write list is complete and
> > send_write_chunks() returns.
> > 
> > That will remain a valid assumption until the server Upper Layer can
> > support multiple bulk payload results per RPC.
> > 
> > Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> > ---
> > net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> > 1 file changed, 7 insertions(+)
> > 
> > diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > index 1dfae83..6c48901 100644
> > --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> > 						arg_ch->rs_handle,
> > 						arg_ch->rs_offset,
> > 						write_len);
> > +
> > +		/* Do not send XDR pad bytes */
> > +		if (chunk_no && write_len < 4) {
> > +			chunk_no++;
> > +			break;
> > +		}
> > +
> > 		chunk_off = 0;
> > 		while (write_len) {
> > 			ret = send_write(xprt, rqstp,
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> --
> Chuck Lever
> 
> 
> 
--
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] 18+ messages in thread

* Re: Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 15:17       ` J. Bruce Fields
  0 siblings, 0 replies; 18+ messages in thread
From: J. Bruce Fields @ 2015-11-12 15:17 UTC (permalink / raw)
  To: Chuck Lever; +Cc: Linux RDMA Mailing List, Linux NFS Mailing List

On Thu, Nov 12, 2015 at 09:47:13AM -0500, Chuck Lever wrote:
> I wish git send-mail had an address book. I’m really tired
> of misspelling the to/cc addresses on patches.

Hah.  Yeah, my hack is a text file of addresses that I cut-and-paste
from.

--b.

> 
> Resending so there is one thread for all replies. Apologies
> for the noise.
> 
> 
> > Begin forwarded message:
> > 
> > From: Chuck Lever <chuck.lever@oracle.com>
> > Subject: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
> > Date: November 12, 2015 at 9:44:33 AM EST
> > To: bfields@fieldses.org
> > Cc: linux-nfs@vger.kernel.org, linux-rmda@vger.kernel.org
> > 
> > Minor optimization: when dealing with write chunk XDR roundup, do
> > not post a Write WR for the zero bytes in the pad. Simply update
> > the write segment in the RPC-over-RDMA header to reflect the extra
> > pad bytes.
> > 
> > The Reply chunk is also a write chunk, but the server does not use
> > send_write_chunks() to send the Reply chunk. That's OK in this case:
> > the server Upper Layer typically marshals the Reply chunk contents
> > in a single contiguous buffer, without a separate tail for the XDR
> > pad.
> > 
> > The comments and the variable naming refer to "chunks" but what is
> > really meant is "segments." The existing code sends only one
> > xdr_write_chunk per RPC reply.
> > 
> > The fix assumes this as well. When the XDR pad in the first write
> > chunk is reached, the assumption is the Write list is complete and
> > send_write_chunks() returns.
> > 
> > That will remain a valid assumption until the server Upper Layer can
> > support multiple bulk payload results per RPC.
> > 
> > Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> > ---
> > net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> > 1 file changed, 7 insertions(+)
> > 
> > diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > index 1dfae83..6c48901 100644
> > --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> > @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> > 						arg_ch->rs_handle,
> > 						arg_ch->rs_offset,
> > 						write_len);
> > +
> > +		/* Do not send XDR pad bytes */
> > +		if (chunk_no && write_len < 4) {
> > +			chunk_no++;
> > +			break;
> > +		}
> > +
> > 		chunk_off = 0;
> > 		while (write_len) {
> > 			ret = send_write(xprt, rqstp,
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> --
> Chuck Lever
> 
> 
> 

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-12 14:47   ` Chuck Lever
@ 2015-11-12 18:30       ` Jason Gunthorpe
  -1 siblings, 0 replies; 18+ messages in thread
From: Jason Gunthorpe @ 2015-11-12 18:30 UTC (permalink / raw)
  To: Chuck Lever
  Cc: J. Bruce Fields, Linux RDMA Mailing List, Linux NFS Mailing List

On Thu, Nov 12, 2015 at 09:47:13AM -0500, Chuck Lever wrote:
> I wish git send-mail had an address book. I’m really tired
> of misspelling the to/cc addresses on patches.

It does:

CONFIGURATION
       sendemail.aliasesfile
           To avoid typing long email addresses, point this to one or more email aliases files. You must also supply
           sendemail.aliasfiletype.

       sendemail.aliasfiletype
           Format of the file(s) specified in sendemail.aliasesfile. Must be one of mutt, mailrc, pine, elm, or gnus.

Jason
--
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] 18+ messages in thread

* Re: Fwd: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 18:30       ` Jason Gunthorpe
  0 siblings, 0 replies; 18+ messages in thread
From: Jason Gunthorpe @ 2015-11-12 18:30 UTC (permalink / raw)
  To: Chuck Lever
  Cc: J. Bruce Fields, Linux RDMA Mailing List, Linux NFS Mailing List

On Thu, Nov 12, 2015 at 09:47:13AM -0500, Chuck Lever wrote:
> I wish git send-mail had an address book. I’m really tired
> of misspelling the to/cc addresses on patches.

It does:

CONFIGURATION
       sendemail.aliasesfile
           To avoid typing long email addresses, point this to one or more email aliases files. You must also supply
           sendemail.aliasfiletype.

       sendemail.aliasfiletype
           Format of the file(s) specified in sendemail.aliasesfile. Must be one of mutt, mailrc, pine, elm, or gnus.

Jason

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-12 14:44 [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk Chuck Lever
  2015-11-12 14:47   ` Chuck Lever
@ 2015-11-12 19:13 ` J. Bruce Fields
  1 sibling, 0 replies; 18+ messages in thread
From: J. Bruce Fields @ 2015-11-12 19:13 UTC (permalink / raw)
  To: Chuck Lever; +Cc: linux-nfs, linux-rmda

Thanks, applying.

On Thu, Nov 12, 2015 at 09:44:33AM -0500, Chuck Lever wrote:
> Minor optimization:

Based on that I'm queueing it up for 4.5--let me know if that causes
some problem.

--b.

> when dealing with write chunk XDR roundup, do
> not post a Write WR for the zero bytes in the pad. Simply update
> the write segment in the RPC-over-RDMA header to reflect the extra
> pad bytes.
> 
> The Reply chunk is also a write chunk, but the server does not use
> send_write_chunks() to send the Reply chunk. That's OK in this case:
> the server Upper Layer typically marshals the Reply chunk contents
> in a single contiguous buffer, without a separate tail for the XDR
> pad.
> 
> The comments and the variable naming refer to "chunks" but what is
> really meant is "segments." The existing code sends only one
> xdr_write_chunk per RPC reply.
> 
> The fix assumes this as well. When the XDR pad in the first write
> chunk is reached, the assumption is the Write list is complete and
> send_write_chunks() returns.
> 
> That will remain a valid assumption until the server Upper Layer can
> support multiple bulk payload results per RPC.
> 
> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> ---
>  net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> index 1dfae83..6c48901 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
>  						arg_ch->rs_handle,
>  						arg_ch->rs_offset,
>  						write_len);
> +
> +		/* Do not send XDR pad bytes */
> +		if (chunk_no && write_len < 4) {
> +			chunk_no++;
> +			break;
> +		}
> +
>  		chunk_off = 0;
>  		while (write_len) {
>  			ret = send_write(xprt, rqstp,

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-12 14:19             ` J. Bruce Fields
@ 2015-11-12 14:31                 ` Chuck Lever
  -1 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-12 14:31 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List


> On Nov 12, 2015, at 9:19 AM, J. Bruce Fields <bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org> wrote:
> 
> On Wed, Nov 11, 2015 at 03:20:33PM -0500, Chuck Lever wrote:
>> 
>>> On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
>>> 
>>> 
>>>> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
>>>> 
>>>> Minor optimization: when dealing with write chunk XDR roundup, do
>>>> not post a Write WR for the zero bytes in the pad. Simply update
>>>> the write segment in the RPC-over-RDMA header to reflect the extra
>>>> pad bytes.
>>>> 
>>>> The Reply chunk is also a write chunk, but the server does not use
>>>> send_write_chunks() to send the Reply chunk. That's OK in this case:
>>>> the server Upper Layer typically marshals the Reply chunk contents
>>>> in a single contiguous buffer, without a separate tail for the XDR
>>>> pad.
>>>> 
>>>> The comments and the variable naming refer to "chunks" but what is
>>>> really meant is "segments." The existing code sends only one
>>>> xdr_write_chunk per RPC reply.
>>>> 
>>>> The fix assumes this as well. When the XDR pad in the first write
>>>> chunk is reached, the assumption is the Write list is complete and
>>>> send_write_chunks() returns.
>>>> 
>>>> That will remain a valid assumption until the server Upper Layer can
>>>> support multiple bulk payload results per RPC.
>>>> 
>>>> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
>>>> --
>>> 
>>> Bruce, can you take this for 4.4 ?
>> 
>> Hi Bruce-
>> 
>> Your 4.4 pull request did not include this patch.
> 
> Whoops, sorry.
> 
> For some reason I don't have the original email, though I do see your
> first followup.  Would you mind resending?
> 
> Also, it looks like that wasn't actually sent to me?  I really want
> submitted patches to have my address on the To: line.  Obviously I also
> try to pick up stuff sent only to the mailing list, but I'm a little
> more likely to miss those.

Yep, my bad, I meant to address that To: you. Will repost.


> But my fault either way, apologies.
> 
> --b.
> 
>> 
>> 
>>>> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
>>>> 1 files changed, 7 insertions(+), 0 deletions(-)
>>>> 
>>>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> index 1dfae83..6c48901 100644
>>>> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
>>>> 						arg_ch->rs_handle,
>>>> 						arg_ch->rs_offset,
>>>> 						write_len);
>>>> +
>>>> +		/* Do not send XDR pad bytes */
>>>> +		if (chunk_no && write_len < 4) {
>>>> +			chunk_no++;
>>>> +			break;
>>>> +		}
>>>> +
>>>> 		chunk_off = 0;
>>>> 		while (write_len) {
>>>> 			ret = send_write(xprt, rqstp,
>>>> 
>>>> --
>>>> 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
>>> 
>>> —
>>> Chuck Lever
>>> 
>>> 
>>> 
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
>>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> 
>> —
>> Chuck Lever
>> 
>> 

—
Chuck Lever



--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" 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] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 14:31                 ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-12 14:31 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List


> On Nov 12, 2015, at 9:19 AM, J. Bruce Fields <bfields@fieldses.org> wrote:
> 
> On Wed, Nov 11, 2015 at 03:20:33PM -0500, Chuck Lever wrote:
>> 
>>> On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever@oracle.com> wrote:
>>> 
>>> 
>>>> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever@oracle.com> wrote:
>>>> 
>>>> Minor optimization: when dealing with write chunk XDR roundup, do
>>>> not post a Write WR for the zero bytes in the pad. Simply update
>>>> the write segment in the RPC-over-RDMA header to reflect the extra
>>>> pad bytes.
>>>> 
>>>> The Reply chunk is also a write chunk, but the server does not use
>>>> send_write_chunks() to send the Reply chunk. That's OK in this case:
>>>> the server Upper Layer typically marshals the Reply chunk contents
>>>> in a single contiguous buffer, without a separate tail for the XDR
>>>> pad.
>>>> 
>>>> The comments and the variable naming refer to "chunks" but what is
>>>> really meant is "segments." The existing code sends only one
>>>> xdr_write_chunk per RPC reply.
>>>> 
>>>> The fix assumes this as well. When the XDR pad in the first write
>>>> chunk is reached, the assumption is the Write list is complete and
>>>> send_write_chunks() returns.
>>>> 
>>>> That will remain a valid assumption until the server Upper Layer can
>>>> support multiple bulk payload results per RPC.
>>>> 
>>>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
>>>> --
>>> 
>>> Bruce, can you take this for 4.4 ?
>> 
>> Hi Bruce-
>> 
>> Your 4.4 pull request did not include this patch.
> 
> Whoops, sorry.
> 
> For some reason I don't have the original email, though I do see your
> first followup.  Would you mind resending?
> 
> Also, it looks like that wasn't actually sent to me?  I really want
> submitted patches to have my address on the To: line.  Obviously I also
> try to pick up stuff sent only to the mailing list, but I'm a little
> more likely to miss those.

Yep, my bad, I meant to address that To: you. Will repost.


> But my fault either way, apologies.
> 
> --b.
> 
>> 
>> 
>>>> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
>>>> 1 files changed, 7 insertions(+), 0 deletions(-)
>>>> 
>>>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> index 1dfae83..6c48901 100644
>>>> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>>>> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
>>>> 						arg_ch->rs_handle,
>>>> 						arg_ch->rs_offset,
>>>> 						write_len);
>>>> +
>>>> +		/* Do not send XDR pad bytes */
>>>> +		if (chunk_no && write_len < 4) {
>>>> +			chunk_no++;
>>>> +			break;
>>>> +		}
>>>> +
>>>> 		chunk_off = 0;
>>>> 		while (write_len) {
>>>> 			ret = send_write(xprt, rqstp,
>>>> 
>>>> --
>>>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>>>> the body of a message to majordomo@vger.kernel.org
>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>> 
>>> —
>>> Chuck Lever
>>> 
>>> 
>>> 
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> 
>> —
>> Chuck Lever
>> 
>> 

—
Chuck Lever




^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-11 20:20         ` Chuck Lever
@ 2015-11-12 14:19             ` J. Bruce Fields
  -1 siblings, 0 replies; 18+ messages in thread
From: J. Bruce Fields @ 2015-11-12 14:19 UTC (permalink / raw)
  To: Chuck Lever; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List

On Wed, Nov 11, 2015 at 03:20:33PM -0500, Chuck Lever wrote:
> 
> > On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
> > 
> > 
> >> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
> >> 
> >> Minor optimization: when dealing with write chunk XDR roundup, do
> >> not post a Write WR for the zero bytes in the pad. Simply update
> >> the write segment in the RPC-over-RDMA header to reflect the extra
> >> pad bytes.
> >> 
> >> The Reply chunk is also a write chunk, but the server does not use
> >> send_write_chunks() to send the Reply chunk. That's OK in this case:
> >> the server Upper Layer typically marshals the Reply chunk contents
> >> in a single contiguous buffer, without a separate tail for the XDR
> >> pad.
> >> 
> >> The comments and the variable naming refer to "chunks" but what is
> >> really meant is "segments." The existing code sends only one
> >> xdr_write_chunk per RPC reply.
> >> 
> >> The fix assumes this as well. When the XDR pad in the first write
> >> chunk is reached, the assumption is the Write list is complete and
> >> send_write_chunks() returns.
> >> 
> >> That will remain a valid assumption until the server Upper Layer can
> >> support multiple bulk payload results per RPC.
> >> 
> >> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> >> --
> > 
> > Bruce, can you take this for 4.4 ?
> 
> Hi Bruce-
> 
> Your 4.4 pull request did not include this patch.

Whoops, sorry.

For some reason I don't have the original email, though I do see your
first followup.  Would you mind resending?

Also, it looks like that wasn't actually sent to me?  I really want
submitted patches to have my address on the To: line.  Obviously I also
try to pick up stuff sent only to the mailing list, but I'm a little
more likely to miss those.

But my fault either way, apologies.

--b.

> 
> 
> >> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> >> 1 files changed, 7 insertions(+), 0 deletions(-)
> >> 
> >> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> index 1dfae83..6c48901 100644
> >> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> >> 						arg_ch->rs_handle,
> >> 						arg_ch->rs_offset,
> >> 						write_len);
> >> +
> >> +		/* Do not send XDR pad bytes */
> >> +		if (chunk_no && write_len < 4) {
> >> +			chunk_no++;
> >> +			break;
> >> +		}
> >> +
> >> 		chunk_off = 0;
> >> 		while (write_len) {
> >> 			ret = send_write(xprt, rqstp,
> >> 
> >> --
> >> 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
> > 
> > —
> > Chuck Lever
> > 
> > 
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> —
> Chuck Lever
> 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" 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] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-12 14:19             ` J. Bruce Fields
  0 siblings, 0 replies; 18+ messages in thread
From: J. Bruce Fields @ 2015-11-12 14:19 UTC (permalink / raw)
  To: Chuck Lever; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List

On Wed, Nov 11, 2015 at 03:20:33PM -0500, Chuck Lever wrote:
> 
> > On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever@oracle.com> wrote:
> > 
> > 
> >> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever@oracle.com> wrote:
> >> 
> >> Minor optimization: when dealing with write chunk XDR roundup, do
> >> not post a Write WR for the zero bytes in the pad. Simply update
> >> the write segment in the RPC-over-RDMA header to reflect the extra
> >> pad bytes.
> >> 
> >> The Reply chunk is also a write chunk, but the server does not use
> >> send_write_chunks() to send the Reply chunk. That's OK in this case:
> >> the server Upper Layer typically marshals the Reply chunk contents
> >> in a single contiguous buffer, without a separate tail for the XDR
> >> pad.
> >> 
> >> The comments and the variable naming refer to "chunks" but what is
> >> really meant is "segments." The existing code sends only one
> >> xdr_write_chunk per RPC reply.
> >> 
> >> The fix assumes this as well. When the XDR pad in the first write
> >> chunk is reached, the assumption is the Write list is complete and
> >> send_write_chunks() returns.
> >> 
> >> That will remain a valid assumption until the server Upper Layer can
> >> support multiple bulk payload results per RPC.
> >> 
> >> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> >> --
> > 
> > Bruce, can you take this for 4.4 ?
> 
> Hi Bruce-
> 
> Your 4.4 pull request did not include this patch.

Whoops, sorry.

For some reason I don't have the original email, though I do see your
first followup.  Would you mind resending?

Also, it looks like that wasn't actually sent to me?  I really want
submitted patches to have my address on the To: line.  Obviously I also
try to pick up stuff sent only to the mailing list, but I'm a little
more likely to miss those.

But my fault either way, apologies.

--b.

> 
> 
> >> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> >> 1 files changed, 7 insertions(+), 0 deletions(-)
> >> 
> >> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> index 1dfae83..6c48901 100644
> >> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> >> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> >> 						arg_ch->rs_handle,
> >> 						arg_ch->rs_offset,
> >> 						write_len);
> >> +
> >> +		/* Do not send XDR pad bytes */
> >> +		if (chunk_no && write_len < 4) {
> >> +			chunk_no++;
> >> +			break;
> >> +		}
> >> +
> >> 		chunk_off = 0;
> >> 		while (write_len) {
> >> 			ret = send_write(xprt, rqstp,
> >> 
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> >> the body of a message to majordomo@vger.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > 
> > —
> > Chuck Lever
> > 
> > 
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> —
> Chuck Lever
> 
> 

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-11-03 20:10     ` Chuck Lever
@ 2015-11-11 20:20         ` Chuck Lever
  -1 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-11 20:20 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List


> On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
> 
> 
>> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
>> 
>> Minor optimization: when dealing with write chunk XDR roundup, do
>> not post a Write WR for the zero bytes in the pad. Simply update
>> the write segment in the RPC-over-RDMA header to reflect the extra
>> pad bytes.
>> 
>> The Reply chunk is also a write chunk, but the server does not use
>> send_write_chunks() to send the Reply chunk. That's OK in this case:
>> the server Upper Layer typically marshals the Reply chunk contents
>> in a single contiguous buffer, without a separate tail for the XDR
>> pad.
>> 
>> The comments and the variable naming refer to "chunks" but what is
>> really meant is "segments." The existing code sends only one
>> xdr_write_chunk per RPC reply.
>> 
>> The fix assumes this as well. When the XDR pad in the first write
>> chunk is reached, the assumption is the Write list is complete and
>> send_write_chunks() returns.
>> 
>> That will remain a valid assumption until the server Upper Layer can
>> support multiple bulk payload results per RPC.
>> 
>> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
>> --
> 
> Bruce, can you take this for 4.4 ?

Hi Bruce-

Your 4.4 pull request did not include this patch.


>> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
>> 1 files changed, 7 insertions(+), 0 deletions(-)
>> 
>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> index 1dfae83..6c48901 100644
>> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
>> 						arg_ch->rs_handle,
>> 						arg_ch->rs_offset,
>> 						write_len);
>> +
>> +		/* Do not send XDR pad bytes */
>> +		if (chunk_no && write_len < 4) {
>> +			chunk_no++;
>> +			break;
>> +		}
>> +
>> 		chunk_off = 0;
>> 		while (write_len) {
>> 			ret = send_write(xprt, rqstp,
>> 
>> --
>> 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
> 
> —
> Chuck Lever
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

—
Chuck Lever



--
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] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-11 20:20         ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-11 20:20 UTC (permalink / raw)
  To: J. Bruce Fields; +Cc: Linux NFS Mailing List, Linux RDMA Mailing List


> On Nov 3, 2015, at 3:10 PM, Chuck Lever <chuck.lever@oracle.com> wrote:
> 
> 
>> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever@oracle.com> wrote:
>> 
>> Minor optimization: when dealing with write chunk XDR roundup, do
>> not post a Write WR for the zero bytes in the pad. Simply update
>> the write segment in the RPC-over-RDMA header to reflect the extra
>> pad bytes.
>> 
>> The Reply chunk is also a write chunk, but the server does not use
>> send_write_chunks() to send the Reply chunk. That's OK in this case:
>> the server Upper Layer typically marshals the Reply chunk contents
>> in a single contiguous buffer, without a separate tail for the XDR
>> pad.
>> 
>> The comments and the variable naming refer to "chunks" but what is
>> really meant is "segments." The existing code sends only one
>> xdr_write_chunk per RPC reply.
>> 
>> The fix assumes this as well. When the XDR pad in the first write
>> chunk is reached, the assumption is the Write list is complete and
>> send_write_chunks() returns.
>> 
>> That will remain a valid assumption until the server Upper Layer can
>> support multiple bulk payload results per RPC.
>> 
>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
>> --
> 
> Bruce, can you take this for 4.4 ?

Hi Bruce-

Your 4.4 pull request did not include this patch.


>> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
>> 1 files changed, 7 insertions(+), 0 deletions(-)
>> 
>> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> index 1dfae83..6c48901 100644
>> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
>> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
>> 						arg_ch->rs_handle,
>> 						arg_ch->rs_offset,
>> 						write_len);
>> +
>> +		/* Do not send XDR pad bytes */
>> +		if (chunk_no && write_len < 4) {
>> +			chunk_no++;
>> +			break;
>> +		}
>> +
>> 		chunk_off = 0;
>> 		while (write_len) {
>> 			ret = send_write(xprt, rqstp,
>> 
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> —
> Chuck Lever
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

—
Chuck Lever




^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
  2015-10-16 13:30 ` Chuck Lever
@ 2015-11-03 20:10     ` Chuck Lever
  -1 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-03 20:10 UTC (permalink / raw)
  To: Linux NFS Mailing List, linux-rdma-u79uwXL29TY76Z2rM5mHXA


> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> wrote:
> 
> Minor optimization: when dealing with write chunk XDR roundup, do
> not post a Write WR for the zero bytes in the pad. Simply update
> the write segment in the RPC-over-RDMA header to reflect the extra
> pad bytes.
> 
> The Reply chunk is also a write chunk, but the server does not use
> send_write_chunks() to send the Reply chunk. That's OK in this case:
> the server Upper Layer typically marshals the Reply chunk contents
> in a single contiguous buffer, without a separate tail for the XDR
> pad.
> 
> The comments and the variable naming refer to "chunks" but what is
> really meant is "segments." The existing code sends only one
> xdr_write_chunk per RPC reply.
> 
> The fix assumes this as well. When the XDR pad in the first write
> chunk is reached, the assumption is the Write list is complete and
> send_write_chunks() returns.
> 
> That will remain a valid assumption until the server Upper Layer can
> support multiple bulk payload results per RPC.
> 
> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> --

Bruce, can you take this for 4.4 ?


> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> 1 files changed, 7 insertions(+), 0 deletions(-)
> 
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> index 1dfae83..6c48901 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> 						arg_ch->rs_handle,
> 						arg_ch->rs_offset,
> 						write_len);
> +
> +		/* Do not send XDR pad bytes */
> +		if (chunk_no && write_len < 4) {
> +			chunk_no++;
> +			break;
> +		}
> +
> 		chunk_off = 0;
> 		while (write_len) {
> 			ret = send_write(xprt, rqstp,
> 
> --
> 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

—
Chuck Lever



--
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] 18+ messages in thread

* Re: [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-11-03 20:10     ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-11-03 20:10 UTC (permalink / raw)
  To: Linux NFS Mailing List, linux-rdma


> On Oct 16, 2015, at 9:30 AM, Chuck Lever <chuck.lever@oracle.com> wrote:
> 
> Minor optimization: when dealing with write chunk XDR roundup, do
> not post a Write WR for the zero bytes in the pad. Simply update
> the write segment in the RPC-over-RDMA header to reflect the extra
> pad bytes.
> 
> The Reply chunk is also a write chunk, but the server does not use
> send_write_chunks() to send the Reply chunk. That's OK in this case:
> the server Upper Layer typically marshals the Reply chunk contents
> in a single contiguous buffer, without a separate tail for the XDR
> pad.
> 
> The comments and the variable naming refer to "chunks" but what is
> really meant is "segments." The existing code sends only one
> xdr_write_chunk per RPC reply.
> 
> The fix assumes this as well. When the XDR pad in the first write
> chunk is reached, the assumption is the Write list is complete and
> send_write_chunks() returns.
> 
> That will remain a valid assumption until the server Upper Layer can
> support multiple bulk payload results per RPC.
> 
> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> --

Bruce, can you take this for 4.4 ?


> net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
> 1 files changed, 7 insertions(+), 0 deletions(-)
> 
> diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> index 1dfae83..6c48901 100644
> --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
> @@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
> 						arg_ch->rs_handle,
> 						arg_ch->rs_offset,
> 						write_len);
> +
> +		/* Do not send XDR pad bytes */
> +		if (chunk_no && write_len < 4) {
> +			chunk_no++;
> +			break;
> +		}
> +
> 		chunk_off = 0;
> 		while (write_len) {
> 			ret = send_write(xprt, rqstp,
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

—
Chuck Lever




^ permalink raw reply	[flat|nested] 18+ messages in thread

* [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-10-16 13:30 ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-10-16 13:30 UTC (permalink / raw)
  To: linux-nfs-u79uwXL29TY76Z2rM5mHXA, linux-rdma-u79uwXL29TY76Z2rM5mHXA

Minor optimization: when dealing with write chunk XDR roundup, do
not post a Write WR for the zero bytes in the pad. Simply update
the write segment in the RPC-over-RDMA header to reflect the extra
pad bytes.

The Reply chunk is also a write chunk, but the server does not use
send_write_chunks() to send the Reply chunk. That's OK in this case:
the server Upper Layer typically marshals the Reply chunk contents
in a single contiguous buffer, without a separate tail for the XDR
pad.

The comments and the variable naming refer to "chunks" but what is
really meant is "segments." The existing code sends only one
xdr_write_chunk per RPC reply.

The fix assumes this as well. When the XDR pad in the first write
chunk is reached, the assumption is the Write list is complete and
send_write_chunks() returns.

That will remain a valid assumption until the server Upper Layer can
support multiple bulk payload results per RPC.

Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
---

 net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)

diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 1dfae83..6c48901 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
 						arg_ch->rs_handle,
 						arg_ch->rs_offset,
 						write_len);
+
+		/* Do not send XDR pad bytes */
+		if (chunk_no && write_len < 4) {
+			chunk_no++;
+			break;
+		}
+
 		chunk_off = 0;
 		while (write_len) {
 			ret = send_write(xprt, rqstp,

--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" 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] 18+ messages in thread

* [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk
@ 2015-10-16 13:30 ` Chuck Lever
  0 siblings, 0 replies; 18+ messages in thread
From: Chuck Lever @ 2015-10-16 13:30 UTC (permalink / raw)
  To: linux-nfs, linux-rdma

Minor optimization: when dealing with write chunk XDR roundup, do
not post a Write WR for the zero bytes in the pad. Simply update
the write segment in the RPC-over-RDMA header to reflect the extra
pad bytes.

The Reply chunk is also a write chunk, but the server does not use
send_write_chunks() to send the Reply chunk. That's OK in this case:
the server Upper Layer typically marshals the Reply chunk contents
in a single contiguous buffer, without a separate tail for the XDR
pad.

The comments and the variable naming refer to "chunks" but what is
really meant is "segments." The existing code sends only one
xdr_write_chunk per RPC reply.

The fix assumes this as well. When the XDR pad in the first write
chunk is reached, the assumption is the Write list is complete and
send_write_chunks() returns.

That will remain a valid assumption until the server Upper Layer can
support multiple bulk payload results per RPC.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---

 net/sunrpc/xprtrdma/svc_rdma_sendto.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)

diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 1dfae83..6c48901 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -342,6 +342,13 @@ static int send_write_chunks(struct svcxprt_rdma *xprt,
 						arg_ch->rs_handle,
 						arg_ch->rs_offset,
 						write_len);
+
+		/* Do not send XDR pad bytes */
+		if (chunk_no && write_len < 4) {
+			chunk_no++;
+			break;
+		}
+
 		chunk_off = 0;
 		while (write_len) {
 			ret = send_write(xprt, rqstp,


^ permalink raw reply related	[flat|nested] 18+ messages in thread

end of thread, other threads:[~2015-11-12 19:13 UTC | newest]

Thread overview: 18+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-11-12 14:44 [PATCH] svcrdma: Do not send XDR roundup bytes for a write chunk Chuck Lever
2015-11-12 14:47 ` Fwd: " Chuck Lever
2015-11-12 14:47   ` Chuck Lever
     [not found]   ` <648A8847-9836-42EC-9D62-77C01C6E9ACF-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-12 15:17     ` J. Bruce Fields
2015-11-12 15:17       ` J. Bruce Fields
2015-11-12 18:30     ` Jason Gunthorpe
2015-11-12 18:30       ` Jason Gunthorpe
2015-11-12 19:13 ` J. Bruce Fields
  -- strict thread matches above, loose matches on Subject: below --
2015-10-16 13:30 Chuck Lever
2015-10-16 13:30 ` Chuck Lever
     [not found] ` <20151016132938.8597.65237.stgit-SEkpj4eus7wUkR7QxMS4uupQh6Td+b7Z@public.gmane.org>
2015-11-03 20:10   ` Chuck Lever
2015-11-03 20:10     ` Chuck Lever
     [not found]     ` <94E65795-E205-44FE-93DE-A1FF24CD5C2E-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-11 20:20       ` Chuck Lever
2015-11-11 20:20         ` Chuck Lever
     [not found]         ` <2CF378F3-09B3-441D-9DE6-5336F0B8BF65-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-11-12 14:19           ` J. Bruce Fields
2015-11-12 14:19             ` J. Bruce Fields
     [not found]             ` <20151112141907.GA6137-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>
2015-11-12 14:31               ` Chuck Lever
2015-11-12 14:31                 ` Chuck Lever

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.