From: "J. Bruce Fields" <bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>
To: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: [PATCH v3 09/18] sunrpc: Allocate one more page per svc_rqst
Date: Thu, 29 Jun 2017 16:20:04 -0400 [thread overview]
Message-ID: <20170629202004.GC4178@fieldses.org> (raw)
In-Reply-To: <20170623211816.5162.54447.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org>
I'm confused by this one:
On Fri, Jun 23, 2017 at 05:18:16PM -0400, Chuck Lever wrote:
> svcrdma needs 259 pages for 1MB NFSv4.0 READ requests:
>
> - 1 page for the transport header and head iovec
> - 256 pages for the data payload
> - 1 page for the trailing GETATTR request (since NFSD XDR decoding
> does not look for a tail iovec, the GETATTR is stuck at the end
> of the rqstp->rq_arg.pages list)
> - 1 page for building the reply xdr_buf
>
> Note that RPCSVC_MAXPAGES is defined as:
>
> ((RPCSVC_MAXPAYLOAD+PAGE_SIZE-1)/PAGE_SIZE + 2 + 1)
>
> I don't understand why the "+PAGE_SIZE-1" is in there, because
> division will always round the result down to
>
> (RPCSVC_MAXPAYLOAD / PAGE_SIZE + 2 + 1)
I think you're assuming that RPCSVC_MAXPAYLOAD is a multiple of
PAGE_SIZE. Maybe that's true in all cases (I haven't checked), but it
seems harmless to handle the case where it's not.
> Let's remove the "-1" to get 260 pages maximum.
Why? (I'm not necessarily opposed, I just missed the explanation.)
--b.
> Then svc_alloc_args() needs to ask for pages according to this
> same formula, otherwise it will never allocate enough.
>
> Signed-off-by: Chuck Lever <chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> ---
> include/linux/sunrpc/svc.h | 3 +--
> net/sunrpc/svc_xprt.c | 8 +++++---
> 2 files changed, 6 insertions(+), 5 deletions(-)
>
> diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
> index 11cef5a..35c274f 100644
> --- a/include/linux/sunrpc/svc.h
> +++ b/include/linux/sunrpc/svc.h
> @@ -176,8 +176,7 @@ static inline void svc_get(struct svc_serv *serv)
> * We using ->sendfile to return read data, we might need one extra page
> * if the request is not page-aligned. So add another '1'.
> */
> -#define RPCSVC_MAXPAGES ((RPCSVC_MAXPAYLOAD+PAGE_SIZE-1)/PAGE_SIZE \
> - + 2 + 1)
> +#define RPCSVC_MAXPAGES ((RPCSVC_MAXPAYLOAD + PAGE_SIZE) / PAGE_SIZE + 2 + 1)
>
> static inline u32 svc_getnl(struct kvec *iov)
> {
> diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
> index 7bfe1fb..9bd484d 100644
> --- a/net/sunrpc/svc_xprt.c
> +++ b/net/sunrpc/svc_xprt.c
> @@ -659,11 +659,13 @@ static int svc_alloc_arg(struct svc_rqst *rqstp)
> int i;
>
> /* now allocate needed pages. If we get a failure, sleep briefly */
> - pages = (serv->sv_max_mesg + PAGE_SIZE) / PAGE_SIZE;
> - WARN_ON_ONCE(pages >= RPCSVC_MAXPAGES);
> - if (pages >= RPCSVC_MAXPAGES)
> + pages = (serv->sv_max_mesg + (2 * PAGE_SIZE)) >> PAGE_SHIFT;
> + if (pages >= RPCSVC_MAXPAGES) {
> + pr_warn_once("svc: warning: pages=%u >= RPCSVC_MAXPAGES=%lu\n",
> + pages, RPCSVC_MAXPAGES);
> /* use as many pages as possible */
> pages = RPCSVC_MAXPAGES - 1;
> + }
> for (i = 0; i < pages ; i++)
> while (rqstp->rq_pages[i] == NULL) {
> struct page *p = alloc_page(GFP_KERNEL);
--
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
next prev parent reply other threads:[~2017-06-29 20:20 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-23 21:16 [PATCH v3 00/18] Server-side NFS/RDMA changes proposed for v4.13 Chuck Lever
[not found] ` <20170623211150.5162.59075.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org>
2017-06-23 21:17 ` [PATCH v3 01/18] sunrpc: Disable splice for krb5i Chuck Lever
2017-06-23 21:17 ` [PATCH v3 02/18] svcrdma: Squelch disconnection messages Chuck Lever
2017-06-23 21:17 ` [PATCH v3 03/18] svcrdma: Avoid Send Queue overflow Chuck Lever
2017-06-23 21:17 ` [PATCH v3 04/18] svcrdma: Remove svc_rdma_marshal.c Chuck Lever
2017-06-23 21:17 ` [PATCH v3 05/18] svcrdma: Improve Read chunk sanity checking Chuck Lever
2017-06-23 21:17 ` [PATCH v3 06/18] svcrdma: Improve Write " Chuck Lever
2017-06-23 21:18 ` [PATCH v3 07/18] svcrdma: Improve Reply " Chuck Lever
2017-06-23 21:18 ` [PATCH v3 08/18] svcrdma: Don't account for Receive queue "starvation" Chuck Lever
2017-06-23 21:18 ` [PATCH v3 09/18] sunrpc: Allocate one more page per svc_rqst Chuck Lever
[not found] ` <20170623211816.5162.54447.stgit-Hs+gFlyCn65vLzlybtyyYzGyq/o6K9yX@public.gmane.org>
2017-06-29 20:20 ` J. Bruce Fields [this message]
[not found] ` <20170629202004.GC4178-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>
2017-06-29 20:44 ` Chuck Lever
[not found] ` <EB8E2100-6176-424A-B687-FE84758889E0-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2017-06-30 0:23 ` Chuck Lever
2017-06-23 21:18 ` [PATCH v3 10/18] svcrdma: Add recvfrom helpers to svc_rdma_rw.c Chuck Lever
2017-06-23 21:18 ` [PATCH v3 11/18] svcrdma: Use generic RDMA R/W API in RPC Call path Chuck Lever
2017-06-23 21:18 ` [PATCH v3 12/18] svcrdma: Properly compute .len and .buflen for received RPC Calls Chuck Lever
2017-06-23 21:18 ` [PATCH v3 13/18] svcrdma: Remove unused Read completion handlers Chuck Lever
2017-06-23 21:18 ` [PATCH v3 14/18] svcrdma: Remove frmr cache Chuck Lever
2017-06-23 21:19 ` [PATCH v3 15/18] svcrdma: Clean-up svc_rdma_unmap_dma Chuck Lever
2017-06-23 21:19 ` [PATCH v3 16/18] svcrdma: Clean up after converting svc_rdma_recvfrom to rdma_rw API Chuck Lever
2017-06-23 21:19 ` [PATCH v3 17/18] svcrdma: use offset_in_page() macro Chuck Lever
2017-06-23 21:19 ` [PATCH v3 18/18] svcrdma: Remove svc_rdma_chunk_ctxt::cc_dir field 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=20170629202004.GC4178@fieldses.org \
--to=bfields-uc3wqj2krung9huczpvpmw@public.gmane.org \
--cc=chuck.lever-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org \
--cc=linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@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