Return-Path: linux-nfs-owner@vger.kernel.org Received: from mail-ie0-f170.google.com ([209.85.223.170]:33942 "EHLO mail-ie0-f170.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755054AbaEOP5E (ORCPT ); Thu, 15 May 2014 11:57:04 -0400 Received: by mail-ie0-f170.google.com with SMTP id ar20so1225735iec.29 for ; Thu, 15 May 2014 08:57:03 -0700 (PDT) From: Weston Andros Adamson To: trond.myklebust@primarydata.com Cc: linux-nfs@vger.kernel.org, Weston Andros Adamson Subject: [PATCH v3 00/18] nfs: support multiple requests per page Date: Thu, 15 May 2014 11:56:39 -0400 Message-Id: <1400169417-28245-1-git-send-email-dros@primarydata.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Sender: linux-nfs-owner@vger.kernel.org List-ID: This patchset changes the read and write paths to be more flexible in dealing with requests that are not page aligned. Until now there was a 1:1 mapping of struct nfs_page (referred to as "nfs requests") to struct page, which limited the client to page aligned I/O in several pNFS scenarios. This patchset allows multiple requests per page, loosely following the approach taken with struct buffer_head (part of kernel bio interface). With this patchset the client now supports: - non-page-aligned O_DIRECT I/O to DSes (instead of reverting to MDS) - arbitrary pnfs layout segment boundaries - arbitrary pnfs filelayout stripe sizes This patchset also includes a lot of cleanup - notably we no longer need a separate code path to support rsize/wsize < PAGE_SIZE. This new approach opens the door to many optimizations, such as not having to flush a page on a non-contiguous write, but for the time being we are focusing on correctness -- this patchset touches the read and write path for *all* versions of NFS! This has been tested against v2, v3, v4.0 and v4.1 (no pnfs) servers with different rsize/wsize settings, and against pynfs filelayout servers hacked to have non page aligned stripe sizes. I had some code review already (with changes applied) and we've been testing this pretty extensively for the last month+ - focusing mostly on v2, v3, v4.x (no pnfs). The patchset applies against Trond's testing branch, but should also include the fix I posted earlier today: "pnfs: fix race in filelayout commit path" as the race seems to be easier to hit with this patchset applied. I'm pretty sure I didn't break anything in the object and block layouts, but some extra attention there would be helpful. I plan on sharing some performance numbers once I'm able to run some nfsometer workloads. Changes in V3: - rebased to Anna's newest patches, which merges pageio.c into pagelist.c Upcoming: - ~5 patch patchset that cleans up nfs_pgio_header and nfs_pgio_data I just pushed an updated version of my ‘pgio’ branch to: git://git.linux-nfs.org/projects/dros/linux-nfs The branch has the following patches: from me (this patchset!): 7d8976b nfs: support page groups in nfs_read_completion f0d7a82 pnfs: filelayout: support non page aligned layouts 79a9c89 pnfs: allow non page aligned pnfs layout segments cf25fb9 pnfs: support multiple verfs per direct req c081af5 nfs: remove data list from pgio header 6313def nfs: use > 1 request to handle bsize < PAGE_SIZE 9f89e40 nfs: chain calls to pg_test d65d153 nfs: allow coalescing of subpage requests 12cf560 pnfs: clean up filelayout_alloc_commit_info 49fa6d6 nfs: page group support in nfs_mark_uptodate ba94674 nfs: page group syncing in write path 00ddd94 nfs: page group syncing in read path d6dec2d nfs: add support for multiple nfs reqs per page 1ea228a nfs: call nfs_can_coalesce_requests for every req 104ab5c nfs: modify pg_test interface to return size_t 27f3d0c nfs: remove unused arg from nfs_create_request fc9eeaa nfs: clean up PG_* flags 366d632 pnfs: fix race in filelayout commit path from Anna: 6e35ebf NFS: Create a common nfs_pageio_ops struct 9a0dabd NFS: Create a common generic_pg_pgios() 9a074ee NFS: Create a common multiple_pgios() function a7c7f54 NFS: Create a common initiate_pgio() function fd42750 NFS: Create a generic_pgio function 4003d02 NFS: Create a common pgio_error function c9f494c NFS: Create a common rpcsetup function for reads and writes 1220495 NFS: Create a common rpc_call_ops struct 4e07fcf NFS: Create a common nfs_pgio_result_common function 1986d26 NFS: Create a common pgio_rpc_prepare function 9892bb1 NFS: Create a common rw_header_alloc and rw_header_free function 76c449c NFS: Create a common pgio_alloc and pgio_release function 4fca7e6 NFS: Move the write verifier into the nfs_pgio_header fd279ad NFS: Create a common read and write header struct 59d6da2 NFS: Create a common read and write data struct 9726903 NFS: Create a common results structure for reads and writes 0ec19c2 NFS: Create a common argument structure for reads and writes from Christoph: 140c1e0 nfs: remove ->read_pageio_init from rpc ops 75fd2e5 nfs: remove ->write_pageio_init from rpc ops The diffstat of pgio branch (including patches from Anna and Christoph): fs/nfs/blocklayout/blocklayout.c | 38 ++- fs/nfs/direct.c | 117 +++++++- fs/nfs/internal.h | 33 +-- fs/nfs/nfs2xdr.c | 14 +- fs/nfs/nfs3proc.c | 21 +- fs/nfs/nfs3xdr.c | 16 +- fs/nfs/nfs4_fs.h | 4 +- fs/nfs/nfs4filelayout.c | 182 ++++++------ fs/nfs/nfs4proc.c | 54 ++-- fs/nfs/nfs4trace.h | 8 +- fs/nfs/nfs4xdr.c | 19 +- fs/nfs/objlayout/objio_osd.c | 24 +- fs/nfs/objlayout/objlayout.c | 16 +- fs/nfs/objlayout/objlayout.h | 8 +- fs/nfs/pagelist.c | 617 ++++++++++++++++++++++++++++++++++++--- fs/nfs/pnfs.c | 146 ++++----- fs/nfs/pnfs.h | 30 +- fs/nfs/proc.c | 21 +- fs/nfs/read.c | 414 +++++--------------------- fs/nfs/write.c | 561 ++++++++++------------------------- include/linux/nfs.h | 5 +- include/linux/nfs_fs.h | 2 - include/linux/nfs_page.h | 46 ++- include/linux/nfs_xdr.h | 106 +++---- 24 files changed, 1276 insertions(+), 1226 deletions(-) -dros Weston Andros Adamson (18): pnfs: fix race in filelayout commit path nfs: clean up PG_* flags nfs: remove unused arg from nfs_create_request nfs: modify pg_test interface to return size_t nfs: call nfs_can_coalesce_requests for every req nfs: add support for multiple nfs reqs per page nfs: page group syncing in read path nfs: page group syncing in write path nfs: page group support in nfs_mark_uptodate pnfs: clean up filelayout_alloc_commit_info nfs: allow coalescing of subpage requests nfs: chain calls to pg_test nfs: use > 1 request to handle bsize < PAGE_SIZE nfs: remove data list from pgio header pnfs: support multiple verfs per direct req pnfs: allow non page aligned pnfs layout segments pnfs: filelayout: support non page aligned layouts nfs: support page groups in nfs_read_completion fs/nfs/blocklayout/blocklayout.c | 16 +- fs/nfs/direct.c | 111 ++++++++++- fs/nfs/nfs4filelayout.c | 142 +++++++------- fs/nfs/objlayout/objio_osd.c | 18 +- fs/nfs/pagelist.c | 387 ++++++++++++++++++++++++++------------- fs/nfs/pnfs.c | 77 ++++---- fs/nfs/pnfs.h | 3 +- fs/nfs/read.c | 51 ++++-- fs/nfs/write.c | 119 ++++++++++-- include/linux/nfs.h | 5 +- include/linux/nfs_page.h | 33 ++-- include/linux/nfs_xdr.h | 7 +- 12 files changed, 666 insertions(+), 303 deletions(-) -- 1.8.5.2 (Apple Git-48)