linux-rdma.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Zhu Yanjun <yanjun.zhu@linux.dev>
To: cel@kernel.org, NeilBrown <neil@brown.name>,
	Jeff Layton <jlayton@kernel.org>,
	Olga Kornievskaia <okorniev@redhat.com>,
	Dai Ngo <dai.ngo@oracle.com>, Tom Talpey <tom@talpey.com>,
	Anna Schumaker <anna@kernel.org>
Cc: linux-nfs@vger.kernel.org, linux-rdma@vger.kernel.org,
	Chuck Lever <chuck.lever@oracle.com>
Subject: Re: [PATCH v4 00/14] Allocate payload arrays dynamically
Date: Tue, 29 Apr 2025 15:06:18 +0200	[thread overview]
Message-ID: <25bb4881-f83b-41c7-8924-d760aa63405b@linux.dev> (raw)
In-Reply-To: <20250428193702.5186-1-cel@kernel.org>

On 28.04.25 21:36, cel@kernel.org wrote:
> From: Chuck Lever <chuck.lever@oracle.com>
> 
> In order to make RPCSVC_MAXPAYLOAD larger (or variable in size), we
> need to do something clever with the payload arrays embedded in
> struct svc_rqst and elsewhere.
> 
> My preference is to keep these arrays allocated all the time because
> allocating them on demand increases the risk of a memory allocation
> failure during a large I/O. This is a quick-and-dirty approach that
> might be replaced once NFSD is converted to use large folios.
> 
> The downside of this design choice is that it pins a few pages per
> NFSD thread (and that's the current situation already). But note
> that because RPCSVC_MAXPAGES is 259, each array is just over a page
> in size, making the allocation waste quite a bit of memory beyond
> the end of the array due to power-of-2 allocator round up. This gets
> worse as the MAXPAGES value is doubled or quadrupled.
> 
> This series also addresses similar issues in the socket and RDMA
> transports.
> 
> v4 is "code complete", unless there are new code change requests.
> I'm not convinced that adding XDR pad alignment to svc_reserve()
> is good, but I'm willing to consider it further.
> 
> It turns out there is already a tuneable for the maximum read and
> write size in NFSD:
> 
>    /proc/fs/nfsd/max_block_size

Hi,

Based on the head commit ca91b9500108 Merge tag 
'v6.15-rc4-ksmbd-server-fixes' of git://git.samba.org/ksmbd, I applied 
this patch series.

When I built the kernel, the following error will pop out.
"
In file included from ./arch/x86/include/asm/bug.h:103,
                  from ./include/linux/bug.h:5,
                  from ./arch/x86/include/asm/paravirt.h:19,
                  from ./arch/x86/include/asm/irqflags.h:102,
                  from ./include/linux/irqflags.h:18,
                  from ./include/linux/spinlock.h:59,
                  from ./include/linux/fs_struct.h:6,
                  from fs/nfsd/nfs4proc.c:35:
fs/nfsd/nfs4proc.c: In function ‘nfsd4_write’:
./include/linux/array_size.h:11:38: warning: division ‘sizeof (struct 
kvec *) / sizeof (struct kvec)’ does not compute the number of array 
elements [-Wsizeof-pointer-div]
    11 | #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + 
__must_be_array(arr))
       |                                      ^
./include/asm-generic/bug.h:111:32: note: in definition of macro 
‘WARN_ON_ONCE’
   111 |         int __ret_warn_on = !!(condition);                      \
       |                                ^~~~~~~~~
fs/nfsd/nfs4proc.c:1231:30: note: in expansion of macro ‘ARRAY_SIZE’
  1231 |         WARN_ON_ONCE(nvecs > ARRAY_SIZE(rqstp->rq_vec));
       |                              ^~~~~~~~~~
./include/linux/compiler.h:197:62: error: static assertion failed: "must 
be array"
   197 | #define __BUILD_BUG_ON_ZERO_MSG(e, msg) ((int)sizeof(struct 
{_Static_assert(!(e), msg);}))
       | 
^~~~~~~~~~~~~~
./include/asm-generic/bug.h:111:32: note: in definition of macro 
‘WARN_ON_ONCE’
   111 |         int __ret_warn_on = !!(condition);                      \
       |                                ^~~~~~~~~
./include/linux/compiler.h:202:33: note: in expansion of macro 
‘__BUILD_BUG_ON_ZERO_MSG’
   202 | #define __must_be_array(a) 
__BUILD_BUG_ON_ZERO_MSG(!__is_array(a), \
       |                                 ^~~~~~~~~~~~~~~~~~~~~~~
./include/linux/array_size.h:11:59: note: in expansion of macro 
‘__must_be_array’
    11 | #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + 
__must_be_array(arr))
       | 
^~~~~~~~~~~~~~~
fs/nfsd/nfs4proc.c:1231:30: note: in expansion of macro ‘ARRAY_SIZE’
  1231 |         WARN_ON_ONCE(nvecs > ARRAY_SIZE(rqstp->rq_vec));
       |                              ^~~~~~~~~~
make[4]: *** [scripts/Makefile.build:203: fs/nfsd/nfs4proc.o] Error 1
make[4]: *** Waiting for unfinished jobs....
make[3]: *** [scripts/Makefile.build:461: fs/nfsd] Error 2
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [scripts/Makefile.build:461: fs] Error 2
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [/home/zyanjun/Development/github-linux/Makefile:2011: .] 
Error 2
make: *** [Makefile:248: __sub-make] Error 2
"

The building host is as below:

$ cat /etc/issue.net
Ubuntu 22.04.5 LTS

$ gcc --version
gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

$ uname -a
Linux lb03055 6.8.0-58-generic #60~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC 
Fri Mar 28 16:09:21 UTC 2 x86_64 x86_64 x86_64 GNU/Linux


Zhu Yanjun

> 
> Since there is an existing user space API for this, my initial
> arguments against adding a tuneable are moot. max_block_size should
> be adequate for this purpose, and enabling it to be set to larger
> values should not impact the kernel-user space API in any way.
> 
> Changes since v3:
> * Improved the rdma_rw context count estimate
> * Dropped "NFSD: Remove NFSSVC_MAXBLKSIZE from .pc_xdrressize"
> * Cleaned up the max size macros a bit
> * Completed the implementation of adjustable max_block_size
> 
> Changes since v2:
> * Address Jeff's review comments
> * Address Neil's review comments
> * Start removing a few uses of NFSSVC_MAXBLKSIZE
> 
> Chuck Lever (14):
>    svcrdma: Reduce the number of rdma_rw contexts per-QP
>    sunrpc: Add a helper to derive maxpages from sv_max_mesg
>    sunrpc: Remove backchannel check in svc_init_buffer()
>    sunrpc: Replace the rq_pages array with dynamically-allocated memory
>    sunrpc: Replace the rq_vec array with dynamically-allocated memory
>    sunrpc: Replace the rq_bvec array with dynamically-allocated memory
>    sunrpc: Adjust size of socket's receive page array dynamically
>    svcrdma: Adjust the number of entries in svc_rdma_recv_ctxt::rc_pages
>    svcrdma: Adjust the number of entries in svc_rdma_send_ctxt::sc_pages
>    sunrpc: Remove the RPCSVC_MAXPAGES macro
>    NFSD: Remove NFSD_BUFSIZE
>    NFSD: Remove NFSSVC_MAXBLKSIZE_V2 macro
>    NFSD: Add a "default" block size
>    SUNRPC: Bump the maximum payload size for the server
> 
>   fs/nfsd/nfs4proc.c                       |  2 +-
>   fs/nfsd/nfs4state.c                      |  2 +-
>   fs/nfsd/nfs4xdr.c                        |  2 +-
>   fs/nfsd/nfsd.h                           | 24 ++++-------
>   fs/nfsd/nfsproc.c                        |  4 +-
>   fs/nfsd/nfssvc.c                         |  2 +-
>   fs/nfsd/nfsxdr.c                         |  4 +-
>   fs/nfsd/vfs.c                            |  2 +-
>   include/linux/sunrpc/svc.h               | 45 +++++++++++++--------
>   include/linux/sunrpc/svc_rdma.h          |  6 ++-
>   include/linux/sunrpc/svcsock.h           |  4 +-
>   net/sunrpc/svc.c                         | 51 +++++++++++++++---------
>   net/sunrpc/svc_xprt.c                    | 10 +----
>   net/sunrpc/svcsock.c                     | 15 ++++---
>   net/sunrpc/xprtrdma/svc_rdma_recvfrom.c  |  8 +++-
>   net/sunrpc/xprtrdma/svc_rdma_rw.c        |  2 +-
>   net/sunrpc/xprtrdma/svc_rdma_sendto.c    | 16 ++++++--
>   net/sunrpc/xprtrdma/svc_rdma_transport.c | 14 ++++---
>   18 files changed, 122 insertions(+), 91 deletions(-)
> 


  parent reply	other threads:[~2025-04-29 13:06 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-28 19:36 [PATCH v4 00/14] Allocate payload arrays dynamically cel
2025-04-28 19:36 ` [PATCH v4 01/14] svcrdma: Reduce the number of rdma_rw contexts per-QP cel
2025-05-06 13:08   ` Christoph Hellwig
2025-05-06 13:17     ` Jason Gunthorpe
2025-05-06 13:40       ` Christoph Hellwig
2025-05-06 13:55         ` Jason Gunthorpe
2025-05-06 14:13           ` Chuck Lever
2025-05-06 14:17             ` Jason Gunthorpe
2025-05-06 14:19               ` Chuck Lever
2025-05-06 14:22                 ` Jason Gunthorpe
2025-05-08  8:41                   ` Edward Srouji
2025-05-08 12:43                     ` Jason Gunthorpe
2025-05-10 23:12                       ` Edward Srouji
2025-04-28 19:36 ` [PATCH v4 02/14] sunrpc: Add a helper to derive maxpages from sv_max_mesg cel
2025-05-06 13:10   ` Christoph Hellwig
2025-04-28 19:36 ` [PATCH v4 03/14] sunrpc: Remove backchannel check in svc_init_buffer() cel
2025-05-06 13:11   ` Christoph Hellwig
2025-04-28 19:36 ` [PATCH v4 04/14] sunrpc: Replace the rq_pages array with dynamically-allocated memory cel
2025-04-30  4:53   ` NeilBrown
2025-04-28 19:36 ` [PATCH v4 05/14] sunrpc: Replace the rq_vec " cel
2025-05-06 13:29   ` Christoph Hellwig
2025-05-06 16:31     ` Chuck Lever
2025-05-07  7:34       ` Christoph Hellwig
2025-04-28 19:36 ` [PATCH v4 06/14] sunrpc: Replace the rq_bvec " cel
2025-04-28 19:36 ` [PATCH v4 07/14] sunrpc: Adjust size of socket's receive page array dynamically cel
2025-04-28 19:36 ` [PATCH v4 08/14] svcrdma: Adjust the number of entries in svc_rdma_recv_ctxt::rc_pages cel
2025-05-06 13:31   ` Christoph Hellwig
2025-05-06 15:20     ` Chuck Lever
2025-05-07  7:40       ` Christoph Hellwig
2025-04-28 19:36 ` [PATCH v4 09/14] svcrdma: Adjust the number of entries in svc_rdma_send_ctxt::sc_pages cel
2025-04-28 19:36 ` [PATCH v4 10/14] sunrpc: Remove the RPCSVC_MAXPAGES macro cel
2025-04-28 19:36 ` [PATCH v4 11/14] NFSD: Remove NFSD_BUFSIZE cel
2025-04-28 21:03   ` Jeff Layton
2025-05-06 13:32   ` Christoph Hellwig
2025-04-28 19:37 ` [PATCH v4 12/14] NFSD: Remove NFSSVC_MAXBLKSIZE_V2 macro cel
2025-05-06 13:33   ` Christoph Hellwig
2025-04-28 19:37 ` [PATCH v4 13/14] NFSD: Add a "default" block size cel
2025-04-28 21:07   ` Jeff Layton
2025-04-28 19:37 ` [PATCH v4 14/14] SUNRPC: Bump the maximum payload size for the server cel
2025-04-28 21:08   ` Jeff Layton
2025-04-29 15:44     ` Chuck Lever
2025-05-06 13:34   ` Christoph Hellwig
2025-05-06 13:52     ` Chuck Lever
2025-05-06 13:54       ` Jeff Layton
2025-05-06 13:59         ` Chuck Lever
2025-05-07  7:42       ` Christoph Hellwig
2025-05-07 14:25         ` Chuck Lever
2025-04-29 13:06 ` Zhu Yanjun [this message]
2025-04-29 13:41   ` [PATCH v4 00/14] Allocate payload arrays dynamically Chuck Lever
2025-04-29 13:52     ` Zhu Yanjun
2025-04-30  5:11 ` NeilBrown
2025-04-30 12:45   ` 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=25bb4881-f83b-41c7-8924-d760aa63405b@linux.dev \
    --to=yanjun.zhu@linux.dev \
    --cc=anna@kernel.org \
    --cc=cel@kernel.org \
    --cc=chuck.lever@oracle.com \
    --cc=dai.ngo@oracle.com \
    --cc=jlayton@kernel.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=neil@brown.name \
    --cc=okorniev@redhat.com \
    --cc=tom@talpey.com \
    /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).