From patchwork Tue Nov 2 12:18:42 2010 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: ehrhardt@linux.vnet.ibm.com X-Patchwork-Id: 297182 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by demeter1.kernel.org (8.14.4/8.14.3) with ESMTP id oA2CJLM1002659 for ; Tue, 2 Nov 2010 12:19:21 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751250Ab0KBMSt (ORCPT ); Tue, 2 Nov 2010 08:18:49 -0400 Received: from mtagate4.uk.ibm.com ([194.196.100.164]:43056 "EHLO mtagate4.uk.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751013Ab0KBMSs (ORCPT ); Tue, 2 Nov 2010 08:18:48 -0400 Received: from d06nrmr1806.portsmouth.uk.ibm.com (d06nrmr1806.portsmouth.uk.ibm.com [9.149.39.193]) by mtagate4.uk.ibm.com (8.13.1/8.13.1) with ESMTP id oA2CIiT6006395; Tue, 2 Nov 2010 12:18:44 GMT Received: from d06av12.portsmouth.uk.ibm.com (d06av12.portsmouth.uk.ibm.com [9.149.37.247]) by d06nrmr1806.portsmouth.uk.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id oA2CIj993891326; Tue, 2 Nov 2010 12:18:45 GMT Received: from d06av12.portsmouth.uk.ibm.com (loopback [127.0.0.1]) by d06av12.portsmouth.uk.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id oA2CIhPE013704; Tue, 2 Nov 2010 06:18:43 -0600 Received: from [9.152.222.30] (dyn-9-152-222-30.boeblingen.de.ibm.com [9.152.222.30]) by d06av12.portsmouth.uk.ibm.com (8.14.4/8.13.1/NCO v10.0 AVin) with ESMTP id oA2CIgbK013700 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Tue, 2 Nov 2010 06:18:43 -0600 Message-ID: <4CD001A2.4000408@linux.vnet.ibm.com> Date: Tue, 02 Nov 2010 13:18:42 +0100 From: Christian Ehrhardt User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1.15) Gecko/20101027 Thunderbird/3.0.10 MIME-Version: 1.0 To: "Jeff Moyer" , "Josef Bacik" , "Chris Mason" , "linux-kernel@vger.kernel.org" , linux-btrfs@vger.kernel.org, "linux-fsdevel@vger.kernel.org" Subject: [RFC][PATCH] direct-io: btrfs: avoid splitting dio requests for non-btrfs filesystems Sender: linux-btrfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-btrfs@vger.kernel.org X-Greylist: IP, sender and recipient auto-whitelisted, not delayed by milter-greylist-4.2.3 (demeter1.kernel.org [140.211.167.41]); Tue, 02 Nov 2010 12:19:22 +0000 (UTC) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index c038644..1126185 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -37,6 +37,7 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" @@ -5822,6 +5823,9 @@ static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb, free_extent_state(cached_state); cached_state = NULL; + /* btrfs cannot handle logically non-contiguous requests */ + kiocb_set_separate_meta_reads(iocb); + ret = __blockdev_direct_IO(rw, iocb, inode, BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev, iov, offset, nr_segs, btrfs_get_blocks_direct, NULL, diff --git a/fs/direct-io.c b/fs/direct-io.c index 48d74c7..6d2dcb2 100644 --- a/fs/direct-io.c +++ b/fs/direct-io.c @@ -35,6 +35,7 @@ #include #include #include +#include #include /* @@ -79,6 +80,7 @@ struct dio { sector_t final_block_in_request;/* doesn't change */ unsigned first_block_in_page; /* doesn't change, Used only once */ int boundary; /* prev block is at a boundary */ + int separate_meta_reads; /* separate in I/O submission */ int reap_counter; /* rate limit reaping */ get_block_t *get_block; /* block mapping function */ dio_iodone_t *end_io; /* IO completion function */ @@ -659,7 +661,7 @@ static int dio_send_cur_page(struct dio *dio) * Submit now if the underlying fs is about to perform a * metadata read */ - else if (dio->boundary) + else if (dio->separate_meta_reads && dio->boundary) dio_bio_submit(dio); } @@ -1245,6 +1247,11 @@ __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode, dio->is_async = !is_sync_kiocb(iocb) && !((rw & WRITE) && (end > i_size_read(inode))); + /* + * some filesystems e.g. btrfs need to separate metadata read + */ + dio->separate_meta_reads = kiocb_needs_separate_meta_reads(iocb); + retval = direct_io_worker(rw, iocb, inode, iov, offset, nr_segs, blkbits, get_block, end_io, submit_io, dio); diff --git a/include/linux/aio.h b/include/linux/aio.h index 7a8db41..9ee9c6e 100644 --- a/include/linux/aio.h +++ b/include/linux/aio.h @@ -34,6 +34,8 @@ struct kioctx; /* #define KIF_LOCKED 0 */ #define KIF_KICKED 1 #define KIF_CANCELLED 2 +/* to separate meta reads */ +#define KIF_SEPARATE_META 3 #define kiocbTryLock(iocb) test_and_set_bit(KIF_LOCKED, &(iocb)->ki_flags) #define kiocbTryKick(iocb) test_and_set_bit(KIF_KICKED, &(iocb)->ki_flags) @@ -50,6 +52,9 @@ struct kioctx; #define kiocbIsKicked(iocb) test_bit(KIF_KICKED, &(iocb)->ki_flags) #define kiocbIsCancelled(iocb) test_bit(KIF_CANCELLED, &(iocb)->ki_flags) +#define kiocb_set_separate_meta_reads(iocb) set_bit(KIF_SEPARATE_META, &(iocb)->ki_flags) +#define kiocb_needs_separate_meta_reads(iocb) (KIF_SEPARATE_META & (iocb)->ki_flags) + /* is there a better place to document function pointer methods? */ /** * ki_retry - iocb forward progress callback