Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759814AbZDQXXt (ORCPT ); Fri, 17 Apr 2009 19:23:49 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754677AbZDQXXj (ORCPT ); Fri, 17 Apr 2009 19:23:39 -0400 Received: from casper.infradead.org ([85.118.1.10]:47961 "EHLO casper.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751469AbZDQXXi (ORCPT ); Fri, 17 Apr 2009 19:23:38 -0400 Subject: Re: Q: NFSD readdir in linux-2.6.28 From: David Woodhouse To: Al Viro Cc: hooanon05@yahoo.co.jp, hch@infradead.org, "linux-kernel@vger.kernel.org" , "linux-fsdevel@vger.kernel.org" In-Reply-To: <20090417225306.GO26366@ZenIV.linux.org.uk> References: <8036.1237474444@jrobl> <1237475837.16359.106.camel@macbook.infradead.org> <8913.1237476890@jrobl> <1239960739.3428.33.camel@macbook.infradead.org> <20090417193233.GM26366@ZenIV.linux.org.uk> <1240006620.19059.41.camel@macbook.infradead.org> <20090417224350.GN26366@ZenIV.linux.org.uk> <20090417225306.GO26366@ZenIV.linux.org.uk> Content-Type: text/plain Date: Sat, 18 Apr 2009 00:23:32 +0100 Message-Id: <1240010612.21423.34.camel@macbook.infradead.org> Mime-Version: 1.0 X-Mailer: Evolution 2.26.1 (2.26.1-2.fc11) Content-Transfer-Encoding: 7bit X-SRS-Rewrite: SMTP reverse-path rewritten from by casper.infradead.org See http://www.infradead.org/rpr.html Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5385 Lines: 191 On Fri, 2009-04-17 at 23:53 +0100, Al Viro wrote: > On Fri, Apr 17, 2009 at 11:43:50PM +0100, Al Viro wrote: > > On Fri, Apr 17, 2009 at 11:17:00PM +0100, David Woodhouse wrote: > > > It sounds like the better answer is to just make sure i_mutex is held > > > when nfsd_buffered_readdir() calls back into the provided filldir > > > function (we could do it in the various filldir functions themselves, > > > _if_ they call lookup_one_len(), but I think I prefer it this way -- > > > it's simpler). Patch below for comment. > > > > Umm... I can live with that, assuming that we don't have callbacks > > that take i_mutex themselves. AFAICS, everything we call there is > > either obviously not touching i_mutex or is already called while we > > hold i_mutex elsewhere, but I'd appreciate if somebody actually > > tested that sucker for different versions of protocol... > > BTW, why mess with taking i_mutex inside the inner loop and not > immediately around it? Or, to phrase my last response slightly differently... because I don't like either of these two versions very much... diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index ab93fcf..0523e9f 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -1885,9 +1885,10 @@ static int nfsd_buffered_filldir(void *__buf, const char *name, int namlen, return 0; } -static int nfsd_buffered_readdir(struct file *file, filldir_t func, - struct readdir_cd *cdp, loff_t *offsetp) +static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func, + struct readdir_cd *cdp, loff_t *offsetp) { + struct inode *dir_inode = file->f_path.dentry->d_inode; struct readdir_data buf; struct buffered_dirent *de; int host_err; @@ -1896,7 +1897,7 @@ static int nfsd_buffered_readdir(struct file *file, filldir_t func, buf.dirent = (void *)__get_free_page(GFP_KERNEL); if (!buf.dirent) - return -ENOMEM; + return nfserrno(-ENOMEM); offset = *offsetp; @@ -1912,23 +1913,37 @@ static int nfsd_buffered_readdir(struct file *file, filldir_t func, host_err = 0; if (host_err < 0) - break; + goto done; size = buf.used; if (!size) - break; + goto done; de = (struct buffered_dirent *)buf.dirent; + + host_err = mutex_lock_killable(&dir_inode->i_mutex); + if (host_err) + goto done; + while (size > 0) { + + int finished; + offset = de->offset; + /* + * 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. + */ + if (func(cdp, de->name, de->namlen, de->offset, de->ino, de->d_type)) - goto done; - + goto done_unlock; + if (cdp->err != nfs_ok) - goto done; + goto done_unlock; reclen = ALIGN(sizeof(*de) + de->namlen, sizeof(u64)); @@ -1937,6 +1952,8 @@ static int nfsd_buffered_readdir(struct file *file, filldir_t func, } offset = vfs_llseek(file, 0, SEEK_CUR); } + done_unlock: + mutex_unlock(&dir_inode->i_mutex); done: free_page((unsigned long)(buf.dirent)); ==================================================================== diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c index ab93fcf..915a47c 100644 --- a/fs/nfsd/vfs.c +++ b/fs/nfsd/vfs.c @@ -1885,8 +1885,8 @@ static int nfsd_buffered_filldir(void *__buf, const char *name, int namlen, return 0; } -static int nfsd_buffered_readdir(struct file *file, filldir_t func, - struct readdir_cd *cdp, loff_t *offsetp) +static __be32 nfsd_buffered_readdir(struct file *file, filldir_t func, + struct readdir_cd *cdp, loff_t *offsetp) { struct readdir_data buf; struct buffered_dirent *de; @@ -1896,11 +1896,12 @@ static int nfsd_buffered_readdir(struct file *file, filldir_t func, buf.dirent = (void *)__get_free_page(GFP_KERNEL); if (!buf.dirent) - return -ENOMEM; + return nfserrno(-ENOMEM); offset = *offsetp; while (1) { + struct inode *dir_inode = file->f_path.dentry->d_inode; unsigned int reclen; cdp->err = nfserr_eof; /* will be cleared on successful read */ @@ -1920,24 +1921,42 @@ static int nfsd_buffered_readdir(struct file *file, filldir_t func, break; de = (struct buffered_dirent *)buf.dirent; + + host_err = mutex_lock_killable(&dir_inode->i_mutex); + if (host_err) + break; + while (size > 0) { + int finished; + offset = de->offset; + /* + * 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. + */ + if (func(cdp, de->name, de->namlen, de->offset, de->ino, de->d_type)) - goto done; - + goto done_unlock; + if (cdp->err != nfs_ok) - goto done; + goto done_unlock; reclen = ALIGN(sizeof(*de) + de->namlen, sizeof(u64)); size -= reclen; de = (struct buffered_dirent *)((char *)de + reclen); } + mutex_unlock(&dir_inode->i_mutex); offset = vfs_llseek(file, 0, SEEK_CUR); - } + continue; + done_unlock: + mutex_unlock(&dir_inode->i_mutex); + break; + } done: free_page((unsigned long)(buf.dirent)); -- dwmw2 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/