From patchwork Wed May 11 10:21:09 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jeff Layton X-Patchwork-Id: 9067481 Return-Path: X-Original-To: patchwork-linux-nfs@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id 462399F1C1 for ; Wed, 11 May 2016 10:22:02 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 1995C20145 for ; Wed, 11 May 2016 10:22:01 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 26C52201B9 for ; Wed, 11 May 2016 10:21:55 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932220AbcEKKVf (ORCPT ); Wed, 11 May 2016 06:21:35 -0400 Received: from mail-qk0-f196.google.com ([209.85.220.196]:35144 "EHLO mail-qk0-f196.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932087AbcEKKV2 (ORCPT ); Wed, 11 May 2016 06:21:28 -0400 Received: by mail-qk0-f196.google.com with SMTP id n62so2438144qkc.2 for ; Wed, 11 May 2016 03:21:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=poochiereds-net.20150623.gappssmtp.com; s=20150623; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=7m0R7ve30Qz74qsG4vC87VVc8qV4eZF3GN/llqDSTUM=; b=gnZAn2PwR/vqBDa1kEpTT6SySb9PZL/bgRRYmpu7aE8poOUgUWknusulAQIlX2hT2y A0mj8rynaOoZjCvlZ2yctiPvPT2lmLkTsTCC41QM1VaaemIcDlXUM5Qt41a08GIG53M0 tn2kCeysH7CZLL1xuEzvNfwVrga4AzDF+EvzsQ1BdduPD6w8wPix9CFNJz4Sx//V6p+i dukzJwfr23v/IUVwS8ktrws0c751QTpwZWdmCLCmBXHCY05iUQ9FtmKuk4V6n4oRdjzG IEhHrm8IK12UKoyPxTc+PmW4SK6Qkqp4M99/T64wBXAjay8MFVeoEQr7CyGdJsJ1Uu9i RnqA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=7m0R7ve30Qz74qsG4vC87VVc8qV4eZF3GN/llqDSTUM=; b=RUr+KhKAI7QcTgDrP6oB1VxLbpD4P9OioWMUVr1CxduH++QZdXDAzcWMKjHnRAOaaG rp6WESHkMjSkLA75t0DBGbNobJZ/+G/KUfOn3bPhGYfUCFFWguHXH1DNG+29HhArJ4K/ Zn/6AUDdu5S6IJTjqtiqZqCwfNYexaB6HCj6QOulNE7OHvh7p2xLXRHdz0WQgB6e2XG+ 7jclaWAqVrW92EeeZv4yQerwqtkCjn08nBTyNz346Qrw1AcGvM/oocalr2+PEggNuSqM X2iIYjuG+2ibeqYH+EiWBF94yW2HTvRv9FMWF+SOeBttryaU2aODhjCPkPPLsJusaTtj MVtQ== X-Gm-Message-State: AOPr4FWZjJprgVp7OMsmtsoObnXtnDHQ7bdX1CmvCn4w1jMLc6jbn5weAD04O03nMvMxQw== X-Received: by 10.55.59.74 with SMTP id i71mr2565145qka.50.1462962087331; Wed, 11 May 2016 03:21:27 -0700 (PDT) Received: from tlielax.poochiereds.net ([2606:a000:1125:4074:3a60:77ff:fe93:a95d]) by smtp.googlemail.com with ESMTPSA id r127sm3048446qkf.47.2016.05.11.03.21.26 (version=TLSv1/SSLv3 cipher=OTHER); Wed, 11 May 2016 03:21:26 -0700 (PDT) From: Jeff Layton X-Google-Original-From: Jeff Layton To: Trond Myklebust , Anna Schumaker Cc: linux-nfs@vger.kernel.org Subject: [PATCH v2 4/9] pnfs: only tear down lsegs that precede seqid in LAYOUTRETURN args Date: Wed, 11 May 2016 06:21:09 -0400 Message-Id: <1462962074-6989-5-git-send-email-jeff.layton@primarydata.com> X-Mailer: git-send-email 2.5.5 In-Reply-To: <1462962074-6989-1-git-send-email-jeff.layton@primarydata.com> References: <1462962074-6989-1-git-send-email-jeff.layton@primarydata.com> Sender: linux-nfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-nfs@vger.kernel.org X-Spam-Status: No, score=-8.2 required=5.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,RP_MATCHES_RCVD,T_DKIM_INVALID,UNPARSEABLE_RELAY autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP LAYOUTRETURN is "special" in that servers and clients are expected to work with old stateids. When the client sends a LAYOUTRETURN with an old stateid in it then the server is expected to only tear down layout segments that were present when that seqid was current. Ensure that the client handles its accounting accordingly. Signed-off-by: Jeff Layton --- fs/nfs/callback_proc.c | 3 ++- fs/nfs/nfs42proc.c | 2 +- fs/nfs/nfs4proc.c | 5 ++-- fs/nfs/pnfs.c | 64 +++++++++++++++++++++++++++++++++----------------- fs/nfs/pnfs.h | 6 +++-- 5 files changed, 52 insertions(+), 28 deletions(-) diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index 618ced381a14..755838df9996 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -217,7 +217,8 @@ static u32 initiate_file_draining(struct nfs_client *clp, } if (pnfs_mark_matching_lsegs_return(lo, &free_me_list, - &args->cbl_range)) { + &args->cbl_range, + be32_to_cpu(args->cbl_stateid.seqid))) { rv = NFS4_OK; goto unlock; } diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c index dff83460e5a6..198bcc3e103d 100644 --- a/fs/nfs/nfs42proc.c +++ b/fs/nfs/nfs42proc.c @@ -232,7 +232,7 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata) * with the current stateid. */ set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags); - pnfs_mark_matching_lsegs_invalid(lo, &head, NULL); + pnfs_mark_matching_lsegs_invalid(lo, &head, NULL, 0); spin_unlock(&inode->i_lock); pnfs_free_lseg_list(&head); } else diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index bc2676c95e1b..c0d75be8cb69 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -7930,7 +7930,7 @@ static void nfs4_layoutget_done(struct rpc_task *task, void *calldata) * with the current stateid. */ set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags); - pnfs_mark_matching_lsegs_invalid(lo, &head, NULL); + pnfs_mark_matching_lsegs_invalid(lo, &head, NULL, 0); spin_unlock(&inode->i_lock); pnfs_free_lseg_list(&head); } else @@ -8122,7 +8122,8 @@ static void nfs4_layoutreturn_release(void *calldata) dprintk("--> %s\n", __func__); spin_lock(&lo->plh_inode->i_lock); - pnfs_mark_matching_lsegs_invalid(lo, &freeme, &lrp->args.range); + pnfs_mark_matching_lsegs_invalid(lo, &freeme, &lrp->args.range, + be32_to_cpu(lrp->args.stateid.seqid)); pnfs_mark_layout_returned_if_empty(lo); if (lrp->res.lrs_present) pnfs_set_layout_stateid(lo, &lrp->res.stateid, true); diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 39432a3705b4..e6cad5ee5d29 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -270,7 +270,7 @@ pnfs_mark_layout_stateid_invalid(struct pnfs_layout_hdr *lo, }; set_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags); - return pnfs_mark_matching_lsegs_invalid(lo, lseg_list, &range); + return pnfs_mark_matching_lsegs_invalid(lo, lseg_list, &range, 0); } static int @@ -308,7 +308,7 @@ pnfs_layout_io_set_failed(struct pnfs_layout_hdr *lo, u32 iomode) spin_lock(&inode->i_lock); pnfs_layout_set_fail_bit(lo, pnfs_iomode_to_fail_bit(iomode)); - pnfs_mark_matching_lsegs_invalid(lo, &head, &range); + pnfs_mark_matching_lsegs_invalid(lo, &head, &range, 0); spin_unlock(&inode->i_lock); pnfs_free_lseg_list(&head); dprintk("%s Setting layout IOMODE_%s fail bit\n", __func__, @@ -522,13 +522,35 @@ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg, return rv; } -/* Returns count of number of matching invalid lsegs remaining in list - * after call. +/* + * Compare 2 layout stateid sequence ids, to see which is newer, + * taking into account wraparound issues. + */ +static bool pnfs_seqid_is_newer(u32 s1, u32 s2) +{ + return (s32)(s1 - s2) > 0; +} + +/** + * pnfs_mark_matching_lsegs_invalid - tear down lsegs or mark them for later + * @lo: layout header containing the lsegs + * @tmp_list: list head where doomed lsegs should go + * @recall_range: optional recall range argument to match (may be NULL) + * @seq: only invalidate lsegs obtained prior to this sequence (may be 0) + * + * Walk the list of lsegs in the layout header, and tear down any that should + * be destroyed. If "recall_range" is specified then the segment must match + * that range. If "seq" is non-zero, then only match segments that were handed + * out at or before that sequence. + * + * Returns number of matching invalid lsegs remaining in list after scanning + * it and purging them. */ int pnfs_mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo, struct list_head *tmp_list, - const struct pnfs_layout_range *recall_range) + const struct pnfs_layout_range *recall_range, + u32 seq) { struct pnfs_layout_segment *lseg, *next; int remaining = 0; @@ -540,10 +562,12 @@ pnfs_mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo, list_for_each_entry_safe(lseg, next, &lo->plh_segs, pls_list) if (!recall_range || should_free_lseg(&lseg->pls_range, recall_range)) { - dprintk("%s: freeing lseg %p iomode %d " + if (seq && pnfs_seqid_is_newer(lseg->pls_seq, seq)) + continue; + dprintk("%s: freeing lseg %p iomode %d seq %u" "offset %llu length %llu\n", __func__, - lseg, lseg->pls_range.iomode, lseg->pls_range.offset, - lseg->pls_range.length); + lseg, lseg->pls_range.iomode, lseg->pls_seq, + lseg->pls_range.offset, lseg->pls_range.length); if (!mark_lseg_invalid(lseg, tmp_list)) remaining++; } @@ -730,15 +754,6 @@ pnfs_destroy_all_layouts(struct nfs_client *clp) pnfs_destroy_layouts_byclid(clp, false); } -/* - * Compare 2 layout stateid sequence ids, to see which is newer, - * taking into account wraparound issues. - */ -static bool pnfs_seqid_is_newer(u32 s1, u32 s2) -{ - return (s32)(s1 - s2) > 0; -} - /* update lo->plh_stateid with new if is more recent */ void pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new, @@ -1014,7 +1029,7 @@ _pnfs_return_layout(struct inode *ino) pnfs_get_layout_hdr(lo); empty = list_empty(&lo->plh_segs); pnfs_clear_layoutcommit(ino, &tmp_list); - pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL); + pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL, 0); if (NFS_SERVER(ino)->pnfs_curr_ld->return_range) { struct pnfs_layout_range range = { @@ -1721,7 +1736,7 @@ pnfs_layout_process(struct nfs4_layoutget *lgp) * inode invalid, and don't bother validating the stateid * sequence number. */ - pnfs_mark_matching_lsegs_invalid(lo, &free_me, NULL); + pnfs_mark_matching_lsegs_invalid(lo, &free_me, NULL, 0); nfs4_stateid_copy(&lo->plh_stateid, &res->stateid); lo->plh_barrier = be32_to_cpu(res->stateid.seqid); @@ -1775,7 +1790,8 @@ pnfs_set_plh_return_info(struct pnfs_layout_hdr *lo, enum pnfs_iomode iomode, int pnfs_mark_matching_lsegs_return(struct pnfs_layout_hdr *lo, struct list_head *tmp_list, - const struct pnfs_layout_range *return_range) + const struct pnfs_layout_range *return_range, + u32 seq) { struct pnfs_layout_segment *lseg, *next; int remaining = 0; @@ -1798,8 +1814,11 @@ pnfs_mark_matching_lsegs_return(struct pnfs_layout_hdr *lo, continue; remaining++; set_bit(NFS_LSEG_LAYOUTRETURN, &lseg->pls_flags); - pnfs_set_plh_return_info(lo, return_range->iomode, lseg->pls_seq); } + + if (remaining) + pnfs_set_plh_return_info(lo, return_range->iomode, seq); + return remaining; } @@ -1822,7 +1841,8 @@ void pnfs_error_mark_layout_for_return(struct inode *inode, * segments at hand when sending layoutreturn. See pnfs_put_lseg() * for how it works. */ - if (!pnfs_mark_matching_lsegs_return(lo, &free_me, &range)) { + if (!pnfs_mark_matching_lsegs_return(lo, &free_me, + &range, lseg->pls_seq)) { nfs4_stateid stateid; enum pnfs_iomode iomode = lo->plh_return_iomode; diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h index 3476c9850678..971068b58647 100644 --- a/fs/nfs/pnfs.h +++ b/fs/nfs/pnfs.h @@ -266,10 +266,12 @@ int pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct nfs4_state *open_state); int pnfs_mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo, struct list_head *tmp_list, - const struct pnfs_layout_range *recall_range); + const struct pnfs_layout_range *recall_range, + u32 seq); int pnfs_mark_matching_lsegs_return(struct pnfs_layout_hdr *lo, struct list_head *tmp_list, - const struct pnfs_layout_range *recall_range); + const struct pnfs_layout_range *recall_range, + u32 seq); bool pnfs_roc(struct inode *ino); void pnfs_roc_release(struct inode *ino); void pnfs_roc_set_barrier(struct inode *ino, u32 barrier);