From: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> To: Sagi Grimberg <sagi-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> Cc: List Linux RDMA Mailing <linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>, Linux NFS Mailing List <linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> Subject: Re: [PATCH v1 03/14] svcrdma: Eliminate RPCRDMA_SQ_DEPTH_MULT Date: Wed, 22 Mar 2017 15:30:22 -0400 [thread overview] Message-ID: <EA949818-1DC4-471C-95C7-C96EA1977236@oracle.com> (raw) In-Reply-To: <68e3eda0-90f4-5bca-28be-b2cf494ed172-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> > On Mar 22, 2017, at 3:06 PM, Sagi Grimberg <sagi-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> wrote: > > >> Roughly speaking, I think there needs to be an rdma_rw API that >> assists the ULP with setting its CQ and SQ sizes, since rdma_rw >> hides the registration mode (one of which, at least, consumes >> more SQEs than the other). > > Hiding the registration mode was the largely the motivation for > this... It buys us simplified implementation and inherently supports > both IB and iWARP (which was annoying and only existing in svc but > still suboptimal). > >> I'd like to introduce one new function call that surfaces the >> factor used to compute how many additional SQEs that rdma_rw will >> need. The ULP will invoke it before allocating new Send CQs. > > I see your point... We should probably get a sense on how to > size the completion queue. I think that this issue is solved with > the CQ pool API that Christoph sent a while ago but was never > pursued. > > The basic idea is that the core would create a pool of long CQs > and then assigns queue-pairs depending on the sq+rq depth. > If we were to pick it up would you consider using it? I will certainly take a look at it. But I don't think that's enough. The ULP is also responsible for managing send queue accounting, and possibly queuing WRs when a send queue is full. So it still needs to know the maximum number of send WRs that can be posted at one time. For svc_rdma, this is sc_sq_avail. I believe that the ULP needs to know the actual number of SQEs both for determining CQ size, and for knowing when to plug the send queue. This maximum depends on the registration mode, the page list depth capability of the HCA (relative to the maximum ULP data payload size), and the page size of the platform. For example, for NFS, the typical maximum rsize and wsize is 1MB. The CX-3 Pro cards I have allow 511 pages per MR in FRWR mode. My systems are x64 using 4KB pages. So I know that one rdma_rw_ctx can handle 256 pages (or 1MB) of payload on my system. An HCA with a smaller page list depth or if the system has larger pages, or an rsize/wsize of 4MB might want a different number of MRs for the same transport, and thus a larger send queue. Alternately, we could set a fixed arbitrary send queue size, and force all ULPs and devices to live with that. That would be much simpler. >> I'll try to provide an RFC in the nfsd-rdma-rw-api topic branch. > > Cool, lets see what you had in mind... -- 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
WARNING: multiple messages have this Message-ID (diff)
From: Chuck Lever <chuck.lever@oracle.com> To: Sagi Grimberg <sagi@grimberg.me> Cc: List Linux RDMA Mailing <linux-rdma@vger.kernel.org>, Linux NFS Mailing List <linux-nfs@vger.kernel.org> Subject: Re: [PATCH v1 03/14] svcrdma: Eliminate RPCRDMA_SQ_DEPTH_MULT Date: Wed, 22 Mar 2017 15:30:22 -0400 [thread overview] Message-ID: <EA949818-1DC4-471C-95C7-C96EA1977236@oracle.com> (raw) In-Reply-To: <68e3eda0-90f4-5bca-28be-b2cf494ed172@grimberg.me> > On Mar 22, 2017, at 3:06 PM, Sagi Grimberg <sagi@grimberg.me> wrote: > > >> Roughly speaking, I think there needs to be an rdma_rw API that >> assists the ULP with setting its CQ and SQ sizes, since rdma_rw >> hides the registration mode (one of which, at least, consumes >> more SQEs than the other). > > Hiding the registration mode was the largely the motivation for > this... It buys us simplified implementation and inherently supports > both IB and iWARP (which was annoying and only existing in svc but > still suboptimal). > >> I'd like to introduce one new function call that surfaces the >> factor used to compute how many additional SQEs that rdma_rw will >> need. The ULP will invoke it before allocating new Send CQs. > > I see your point... We should probably get a sense on how to > size the completion queue. I think that this issue is solved with > the CQ pool API that Christoph sent a while ago but was never > pursued. > > The basic idea is that the core would create a pool of long CQs > and then assigns queue-pairs depending on the sq+rq depth. > If we were to pick it up would you consider using it? I will certainly take a look at it. But I don't think that's enough. The ULP is also responsible for managing send queue accounting, and possibly queuing WRs when a send queue is full. So it still needs to know the maximum number of send WRs that can be posted at one time. For svc_rdma, this is sc_sq_avail. I believe that the ULP needs to know the actual number of SQEs both for determining CQ size, and for knowing when to plug the send queue. This maximum depends on the registration mode, the page list depth capability of the HCA (relative to the maximum ULP data payload size), and the page size of the platform. For example, for NFS, the typical maximum rsize and wsize is 1MB. The CX-3 Pro cards I have allow 511 pages per MR in FRWR mode. My systems are x64 using 4KB pages. So I know that one rdma_rw_ctx can handle 256 pages (or 1MB) of payload on my system. An HCA with a smaller page list depth or if the system has larger pages, or an rsize/wsize of 4MB might want a different number of MRs for the same transport, and thus a larger send queue. Alternately, we could set a fixed arbitrary send queue size, and force all ULPs and devices to live with that. That would be much simpler. >> I'll try to provide an RFC in the nfsd-rdma-rw-api topic branch. > > Cool, lets see what you had in mind... -- Chuck Lever
next prev parent reply other threads:[~2017-03-22 19:30 UTC|newest] Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top 2017-03-16 15:52 [PATCH v1 00/14] Server-side NFS/RDMA changes for v4.12 Chuck Lever 2017-03-16 15:52 ` Chuck Lever [not found] ` <20170316154132.4482.56769.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-16 15:52 ` [PATCH v1 01/14] svcrdma: Move send_wr to svc_rdma_op_ctxt Chuck Lever 2017-03-16 15:52 ` Chuck Lever [not found] ` <20170316155234.4482.94225.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-21 17:49 ` Sagi Grimberg 2017-03-21 17:49 ` Sagi Grimberg 2017-03-16 15:52 ` [PATCH v1 02/14] svcrdma: Add svc_rdma_map_reply_hdr() Chuck Lever 2017-03-16 15:52 ` Chuck Lever [not found] ` <20170316155242.4482.64809.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-21 17:54 ` Sagi Grimberg 2017-03-21 17:54 ` Sagi Grimberg [not found] ` <f5000e25-6ca1-fc24-35c0-6089cf50923c-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> 2017-03-21 18:40 ` Chuck Lever 2017-03-21 18:40 ` Chuck Lever [not found] ` <A18F9D5E-09BA-4268-9AA6-3E5866101F76-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> 2017-03-22 13:07 ` Sagi Grimberg 2017-03-22 13:07 ` Sagi Grimberg 2017-03-16 15:52 ` [PATCH v1 03/14] svcrdma: Eliminate RPCRDMA_SQ_DEPTH_MULT Chuck Lever 2017-03-16 15:52 ` Chuck Lever [not found] ` <20170316155250.4482.49638.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-21 17:58 ` Sagi Grimberg 2017-03-21 17:58 ` Sagi Grimberg [not found] ` <46eb6195-a542-b35c-4902-a2bebb38feba-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> 2017-03-21 18:44 ` Chuck Lever 2017-03-21 18:44 ` Chuck Lever [not found] ` <391F0D90-2A46-4B2F-BCF0-B3BE7D48A3EF-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> 2017-03-22 13:09 ` Sagi Grimberg 2017-03-22 13:09 ` Sagi Grimberg [not found] ` <ec82feb4-d6b9-7fb4-5b11-b8007e313845-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> 2017-03-22 13:36 ` Chuck Lever 2017-03-22 13:36 ` Chuck Lever [not found] ` <C9D8A91C-DE08-4A41-A07D-1F4C42DD9B97-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> 2017-03-22 19:06 ` Sagi Grimberg 2017-03-22 19:06 ` Sagi Grimberg [not found] ` <68e3eda0-90f4-5bca-28be-b2cf494ed172-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> 2017-03-22 19:30 ` Chuck Lever [this message] 2017-03-22 19:30 ` Chuck Lever 2017-03-16 15:52 ` [PATCH v1 04/14] svcrdma: Add helper to save pages under I/O Chuck Lever 2017-03-16 15:52 ` Chuck Lever [not found] ` <20170316155258.4482.69182.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-21 18:01 ` Sagi Grimberg 2017-03-21 18:01 ` Sagi Grimberg 2017-03-16 15:53 ` [PATCH v1 05/14] svcrdma: Introduce local rdma_rw API helpers Chuck Lever 2017-03-16 15:53 ` Chuck Lever [not found] ` <20170316155306.4482.68041.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:17 ` Sagi Grimberg 2017-03-22 14:17 ` Sagi Grimberg [not found] ` <cfa49433-ab26-d2f0-27d4-2a96ff0adaba-NQWnxTmZq1alnMjI0IkVqw@public.gmane.org> 2017-03-22 15:41 ` Chuck Lever 2017-03-22 15:41 ` Chuck Lever [not found] ` <1CAD2542-A121-47ED-A47C-624E188EB54F-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org> 2017-03-24 22:19 ` Chuck Lever 2017-03-24 22:19 ` Chuck Lever 2017-03-16 15:53 ` [PATCH v1 06/14] svcrdma: Use rdma_rw API in RPC reply path Chuck Lever 2017-03-16 15:53 ` Chuck Lever 2017-03-16 15:53 ` [PATCH v1 07/14] svcrdma: Clean up RDMA_ERROR path Chuck Lever 2017-03-16 15:53 ` Chuck Lever [not found] ` <20170316155323.4482.8051.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:18 ` Sagi Grimberg 2017-03-22 14:18 ` Sagi Grimberg 2017-03-16 15:53 ` [PATCH v1 08/14] svcrdma: Report Write/Reply chunk overruns Chuck Lever 2017-03-16 15:53 ` Chuck Lever [not found] ` <20170316155331.4482.7734.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:20 ` Sagi Grimberg 2017-03-22 14:20 ` Sagi Grimberg 2017-03-16 15:53 ` [PATCH v1 09/14] svcrdma: Clean up RPC-over-RDMA backchannel reply processing Chuck Lever 2017-03-16 15:53 ` Chuck Lever 2017-03-16 15:53 ` [PATCH v1 10/14] svcrdma: Reduce size of sge array in struct svc_rdma_op_ctxt Chuck Lever 2017-03-16 15:53 ` Chuck Lever [not found] ` <20170316155347.4482.74652.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:21 ` Sagi Grimberg 2017-03-22 14:21 ` Sagi Grimberg 2017-03-16 15:53 ` [PATCH v1 11/14] svcrdma: Remove old RDMA Write completion handlers Chuck Lever 2017-03-16 15:53 ` Chuck Lever [not found] ` <20170316155355.4482.35026.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:22 ` Sagi Grimberg 2017-03-22 14:22 ` Sagi Grimberg 2017-03-16 15:54 ` [PATCH v1 12/14] svcrdma: Remove the req_map cache Chuck Lever 2017-03-16 15:54 ` Chuck Lever [not found] ` <20170316155403.4482.2040.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:22 ` Sagi Grimberg 2017-03-22 14:22 ` Sagi Grimberg 2017-03-16 15:54 ` [PATCH v1 13/14] svcrdma: Clean out old XDR encoders Chuck Lever 2017-03-16 15:54 ` Chuck Lever [not found] ` <20170316155411.4482.37224.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org> 2017-03-22 14:23 ` Sagi Grimberg 2017-03-22 14:23 ` Sagi Grimberg 2017-03-16 15:54 ` [PATCH v1 14/14] svcrdma: Clean up svc_rdma_post_recv() error handling Chuck Lever 2017-03-16 15:54 ` Chuck Lever
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=EA949818-1DC4-471C-95C7-C96EA1977236@oracle.com \ --to=chuck.lever-qhclzuegtsvqt0dzr+alfa@public.gmane.org \ --cc=linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \ --cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \ --cc=sagi-NQWnxTmZq1alnMjI0IkVqw@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: linkBe 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.