From patchwork Tue Aug 18 19:10:28 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "J. Bruce Fields" X-Patchwork-Id: 7032931 Return-Path: X-Original-To: patchwork-linux-nfs@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork2.web.kernel.org (Postfix) with ESMTP id 49B35C05AC for ; Tue, 18 Aug 2015 19:10:34 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id B26A920656 for ; Tue, 18 Aug 2015 19:10:32 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 45B6D2064B for ; Tue, 18 Aug 2015 19:10:31 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753848AbbHRTK3 (ORCPT ); Tue, 18 Aug 2015 15:10:29 -0400 Received: from fieldses.org ([173.255.197.46]:34982 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753117AbbHRTK2 (ORCPT ); Tue, 18 Aug 2015 15:10:28 -0400 Received: by fieldses.org (Postfix, from userid 2815) id 4789A249A; Tue, 18 Aug 2015 15:10:28 -0400 (EDT) Date: Tue, 18 Aug 2015 15:10:28 -0400 From: "J. Bruce Fields" To: NeilBrown Cc: Kinglong Mee , "linux-nfs@vger.kernel.org" , Al Viro , linux-fsdevel@vger.kernel.org Subject: Re: [PATCH] nfsd: don't hold i_mutex over userspace upcalls Message-ID: <20150818191028.GA3957@fieldses.org> References: <20150430213602.GB9509@fieldses.org> <20150501115326.51f5613a@notabene.brown> <20150504220130.GB16827@fieldses.org> <5548CBA3.9080608@gmail.com> <20150505141801.GB27106@fieldses.org> <20150505155203.GD27106@fieldses.org> <20150506082628.0dd049c7@notabene.brown> <20150508161558.GA18851@fieldses.org> <20150508200133.GC18851@fieldses.org> <20150603151819.GA8441@fieldses.org> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <20150603151819.GA8441@fieldses.org> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-nfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-nfs@vger.kernel.org X-Spam-Status: No, score=-7.3 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, RP_MATCHES_RCVD, 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 Al, can I get an ACK/NACK for this? --b. commit 5494e10316e3 Author: NeilBrown Date: Fri May 1 11:53:26 2015 +1000 nfsd: don't hold i_mutex over userspace upcalls We need information about exports when crossing mountpoints during lookup or NFSv4 readdir. If we don't already have that information cached, we may have to ask (and wait for) rpc.mountd. In both cases we currently hold the i_mutex on the parent of the directory we're asking rpc.mountd about. We've seen situations where rpc.mountd performs some operation on that directory that tries to take the i_mutex again, resulting in deadlock. With some care, we may be able to avoid that in rpc.mountd. But it seems better just to avoid holding a mutex while waiting on userspace. It appears that lookup_one_len is pretty much the only operation that needs the i_mutex. So we could just drop the i_mutex elsewhere and do something like mutex_lock() lookup_one_len() mutex_unlock() In many cases though the lookup would have been cached and not required the i_mutex, so it's more efficient to create a lookup_one_len() variant that only takes the i_mutex when necessary. Signed-off-by: NeilBrown Signed-off-by: J. Bruce Fields --- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/fs/namei.c b/fs/namei.c index ae4e4c18b2ac..1627d0302a59 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -2250,6 +2250,8 @@ EXPORT_SYMBOL(vfs_path_lookup); * * Note that this routine is purely a helper for filesystem usage and should * not be called by generic code. + * + * The caller must hold base->i_mutex. */ struct dentry *lookup_one_len(const char *name, struct dentry *base, int len) { @@ -2293,6 +2295,78 @@ struct dentry *lookup_one_len(const char *name, struct dentry *base, int len) } EXPORT_SYMBOL(lookup_one_len); +/** + * lookup_one_len_unlocked - filesystem helper to lookup single pathname component + * @name: pathname component to lookup + * @base: base directory to lookup from + * @len: maximum length @len should be interpreted to + * + * Note that this routine is purely a helper for filesystem usage and should + * not be called by generic code. + * + * Unlike lookup_one_len, it should be called without the parent + * i_mutex held, and will take the i_mutex itself if necessary. + */ +struct dentry *lookup_one_len_unlocked(const char *name, + struct dentry *base, int len) +{ + struct qstr this; + unsigned int c; + int err; + struct dentry *ret; + + this.name = name; + this.len = len; + this.hash = full_name_hash(name, len); + if (!len) + return ERR_PTR(-EACCES); + + if (unlikely(name[0] == '.')) { + if (len < 2 || (len == 2 && name[1] == '.')) + return ERR_PTR(-EACCES); + } + + while (len--) { + c = *(const unsigned char *)name++; + if (c == '/' || c == '\0') + return ERR_PTR(-EACCES); + } + /* + * See if the low-level filesystem might want + * to use its own hash.. + */ + if (base->d_flags & DCACHE_OP_HASH) { + int err = base->d_op->d_hash(base, &this); + if (err < 0) + return ERR_PTR(err); + } + + err = inode_permission(base->d_inode, MAY_EXEC); + if (err) + return ERR_PTR(err); + + ret = __d_lookup(base, &this); + if (ret) + return ret; + /* + * __d_lookup() is used to try to get a quick answer and avoid the + * mutex. A false-negative does no harm. + */ + ret = __d_lookup(base, &this); + if (ret && ret->d_flags & DCACHE_OP_REVALIDATE) { + dput(ret); + ret = NULL; + } + if (ret) + return ret; + + mutex_lock(&base->d_inode->i_mutex); + ret = __lookup_hash(&this, base, 0); + mutex_unlock(&base->d_inode->i_mutex); + return ret; +} +EXPORT_SYMBOL(lookup_one_len_unlocked); + int user_path_at_empty(int dfd, const char __user *name, unsigned flags, struct path *path, int *empty) { diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c index f6e7cbabac5a..01dcd494f781 100644 --- a/fs/nfsd/nfs3xdr.c +++ b/fs/nfsd/nfs3xdr.c @@ -823,7 +823,7 @@ compose_entry_fh(struct nfsd3_readdirres *cd, struct svc_fh *fhp, } else dchild = dget(dparent); } else - dchild = lookup_one_len(name, dparent, namlen); + dchild = lookup_one_len_unlocked(name, dparent, namlen); if (IS_ERR(dchild)) return rv; if (d_mountpoint(dchild)) diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c index 51c9e9ca39a4..325521ce389a 100644 --- a/fs/nfsd/nfs4xdr.c +++ b/fs/nfsd/nfs4xdr.c @@ -2838,14 +2838,14 @@ nfsd4_encode_dirent_fattr(struct xdr_stream *xdr, struct nfsd4_readdir *cd, __be32 nfserr; int ignore_crossmnt = 0; - dentry = lookup_one_len(name, cd->rd_fhp->fh_dentry, namlen); + dentry = lookup_one_len_unlocked(name, cd->rd_fhp->fh_dentry, namlen); if (IS_ERR(dentry)) return nfserrno(PTR_ERR(dentry)); if (d_really_is_negative(dentry)) { /* - * nfsd_buffered_readdir drops the i_mutex between - * readdir and calling this callback, leaving a window - * where this directory entry could have gone away. + * we're not holding the i_mutex here, so there's + * a window where this directory entry could have gone + * away. */ dput(dentry); return nfserr_noent; diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index 45c04979e7b3..2ea6c6a37364 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -217,10 +217,16 @@ nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp, host_err = PTR_ERR(dentry); if (IS_ERR(dentry)) goto out_nfserr; - /* - * check if we have crossed a mount point ... - */ if (nfsd_mountpoint(dentry, exp)) { + /* + * We don't need the i_mutex after all. It's + * still possible we could open this (regular + * files can be mountpoints too), but the + * i_mutex is just there to prevent renames of + * something that we might be about to delegate, + * and a mountpoint won't be renamed: + */ + fh_unlock(fhp); if ((host_err = nfsd_cross_mnt(rqstp, &dentry, &exp))) { dput(dentry); goto out_nfserr; @@ -1809,7 +1815,6 @@ static __be32 nfsd_buffered_readdir(struct file *file, nfsd_filldir_t func, offset = *offsetp; while (1) { - struct inode *dir_inode = file_inode(file); unsigned int reclen; cdp->err = nfserr_eof; /* will be cleared on successful read */ @@ -1828,15 +1833,6 @@ static __be32 nfsd_buffered_readdir(struct file *file, nfsd_filldir_t func, if (!size) break; - /* - * Various filldir functions may end up calling back into - * lookup_one_len() and the file system's ->lookup() method. - * These expect i_mutex to be held, as it would within readdir. - */ - host_err = mutex_lock_killable(&dir_inode->i_mutex); - if (host_err) - break; - de = (struct buffered_dirent *)buf.dirent; while (size > 0) { offset = de->offset; @@ -1853,7 +1849,6 @@ static __be32 nfsd_buffered_readdir(struct file *file, nfsd_filldir_t func, size -= reclen; de = (struct buffered_dirent *)((char *)de + reclen); } - mutex_unlock(&dir_inode->i_mutex); if (size > 0) /* We bailed out early */ break; diff --git a/include/linux/namei.h b/include/linux/namei.h index d8c6334cd150..d0f25d81b46a 100644 --- a/include/linux/namei.h +++ b/include/linux/namei.h @@ -77,6 +77,7 @@ extern struct dentry *kern_path_locked(const char *, struct path *); extern int kern_path_mountpoint(int, const char *, struct path *, unsigned int); extern struct dentry *lookup_one_len(const char *, struct dentry *, int); +extern struct dentry *lookup_one_len_unlocked(const char *, struct dentry *, int); extern int follow_down_one(struct path *); extern int follow_down(struct path *);