linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tao Ma <tao.ma@oracle.com>
To: Tejun Heo <tj@kernel.org>
Cc: linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH 02/41] block/loop: queue ordered mode should be DRAIN_FLUSH
Date: Sun, 12 Sep 2010 19:55:47 +0800	[thread overview]
Message-ID: <4C8CBFC3.8060706@oracle.com> (raw)
In-Reply-To: <4C8CBC50.1070400@kernel.org>



On 09/12/2010 07:41 PM, Tejun Heo wrote:
> Hello,
>
> On 09/12/2010 10:38 AM, Tao Ma wrote:
>> According to Documentation/block/barrier.txt,
>> QUEUE_ORDERED_DRAIN
>>          Requests are ordered by draining the request queue and cache
>>          flushing isn't needed.
>>
>>          Sequence: drain =>  barrier
>>
>> QUEUE_ORDERED_DRAIN_FLUSH
>>          Requests are ordered by draining the request queue and both
>>          pre-barrier and post-barrier cache flushings are needed.
>>
>>          Sequence: drain =>  preflush =>  barrier =>  postflush
>>
>> And for loop device, it call fsync in barrier request. See do_bio_filebacked in drivers/block/loop.c
>> bool barrier = !!(bio->bi_rw&  REQ_HARDBARRIER);
>>                  struct file *file = lo->lo_backing_file;
>>
>>                  if (barrier) {
>>                          if (unlikely(!file->f_op->fsync)) {
>>                                  ret = -EOPNOTSUPP;
>>                                  goto out;
>>                          }
>>
>>                          ret = vfs_fsync(file, 0);
>>                          if (unlikely(ret)) {
>>                                  ret = -EIO;
>>                                  goto out;
>>                          }
>>                  }
>>
>> So actually the sync is done in barrier and no extra flush is needed?
>> Or Do I misread the rules in barrier.txt?
>
> Hmmm... maybe the doc was a bit confusing.  Any device which has
> writeback cache should have FLUSH in the queue ordered configuration.
> The loop device used vfs_fsync() for cache flushing and didn't support
> ordered sequence.  As such, it should use draining for request
> ordering and suports FLUSH, so the mode to use is DRAIN_FLUSH.
oh, thanks for the explanation.

Regards,
Tao

  reply	other threads:[~2010-09-12 11:56 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-09-03 10:29 [PATCHSET #upstream] block, fs: replace HARDBARRIER with FLUSH/FUA, take#2 Tejun Heo
2010-09-03 10:29 ` [PATCH 01/41] ide: remove unnecessary blk_queue_flushing() test in do_ide_request() Tejun Heo
2010-09-03 10:29 ` [PATCH 02/41] block/loop: queue ordered mode should be DRAIN_FLUSH Tejun Heo
2010-09-12  8:38   ` Tao Ma
2010-09-12 11:41     ` Tejun Heo
2010-09-12 11:55       ` Tao Ma [this message]
2010-09-03 10:29 ` [PATCH 03/41] block: kill QUEUE_ORDERED_BY_TAG Tejun Heo
2010-09-03 10:29 ` [PATCH 04/41] block: deprecate barrier and replace blk_queue_ordered() with blk_queue_flush() Tejun Heo
2010-09-03 10:29 ` [PATCH 05/41] block: remove spurious uses of REQ_HARDBARRIER Tejun Heo
2010-09-03 10:29 ` [PATCH 06/41] block: misc cleanups in barrier code Tejun Heo
2010-09-03 10:29 ` [PATCH 07/41] block: drop barrier ordering by queue draining Tejun Heo
2010-09-03 10:29 ` [PATCH 08/41] block: rename blk-barrier.c to blk-flush.c Tejun Heo
2010-09-03 10:29 ` [PATCH 09/41] block: rename barrier/ordered to flush Tejun Heo
2010-09-03 10:29 ` [PATCH 10/41] block: implement REQ_FLUSH/FUA based interface for FLUSH/FUA requests Tejun Heo
2010-09-03 10:29 ` [PATCH 11/41] block: filter flush bio's in __generic_make_request() Tejun Heo
2010-09-03 10:29 ` [PATCH 12/41] block: simplify queue_next_fseq Tejun Heo
2010-09-03 10:29 ` [PATCH 13/41] block: initialize flush request with WRITE_FLUSH instead of REQ_FLUSH Tejun Heo
2010-09-03 10:29 ` [PATCH 14/41] block: kick queue after sequencing REQ_FLUSH/FUA Tejun Heo
2010-09-03 10:29 ` [PATCH 15/41] block: make sure FSEQ_DATA request has the same rq_disk as the original Tejun Heo
2010-09-03 10:29 ` [PATCH 16/41] block: update documentation for REQ_FLUSH / REQ_FUA Tejun Heo
2010-09-03 10:29 ` [PATCH 17/41] block: use REQ_FLUSH in blkdev_issue_flush() Tejun Heo
2010-09-03 10:29 ` [PATCH 18/41] block/loop: implement REQ_FLUSH/FUA support Tejun Heo
2010-09-03 10:29 ` [PATCH 19/41] virtio_blk: drop REQ_HARDBARRIER support Tejun Heo
2010-09-03 10:29 ` [PATCH 20/41] lguest: replace VIRTIO_F_BARRIER support with VIRTIO_F_FLUSH support Tejun Heo
2010-09-03 10:29 ` [PATCH 21/41] md: implment REQ_FLUSH/FUA support Tejun Heo
2010-09-03 10:29 ` [PATCH 22/41] block: make __blk_rq_prep_clone() copy most command flags Tejun Heo
2010-09-03 10:29 ` [PATCH 23/41] dm: implement REQ_FLUSH/FUA support for bio-based dm Tejun Heo
2010-09-03 12:36   ` Mike Snitzer
2010-09-06 11:14   ` [dm-devel] " Milan Broz
2010-09-07 21:17     ` Mike Snitzer
2010-09-07 22:15       ` Mike Snitzer
2010-09-07 23:49         ` [PATCH 42/41] dm: convey that all flushes are processed as empty Mike Snitzer
2010-09-08  0:00           ` Christoph Hellwig
2010-09-08  2:04             ` [PATCH 42/41 v2] " Mike Snitzer
2010-09-08 16:09               ` Tejun Heo
2010-09-08 16:09               ` Tejun Heo
2010-09-10 18:25   ` [PATCH 23/41] dm: implement REQ_FLUSH/FUA support for bio-based dm Mikulas Patocka
2010-09-10 18:46     ` Mike Snitzer
2010-09-10 19:05       ` Mikulas Patocka
2010-09-10 19:24         ` Mike Snitzer
2010-09-10 20:06           ` Mikulas Patocka
2010-09-10 23:36             ` Tejun Heo
2010-09-11  1:46               ` Mike Snitzer
2010-09-18 17:58                 ` Bill Davidsen
2010-09-18 20:42                   ` [dm-devel] " Mike Snitzer
2010-09-11 12:19               ` Ric Wheeler
2010-09-13 19:01               ` Mikulas Patocka
2010-09-03 10:29 ` [PATCH 24/41] dm: implement REQ_FLUSH/FUA support for request-based dm Tejun Heo
2010-09-08  1:46   ` Kiyoshi Ueda
2010-09-03 10:29 ` [PATCH 25/41] dm: relax ordering of bio-based flush implementation Tejun Heo
2010-09-03 10:29 ` [PATCH 26/41] dm: fix locking context in queue_io() Tejun Heo
2010-09-03 10:29 ` [PATCH 27/41] block: pass gfp_mask and flags to sb_issue_discard Tejun Heo
2010-09-03 10:29 ` [PATCH 28/41] xfs: replace barriers with explicit flush / FUA usage Tejun Heo
2010-09-03 10:29 ` [PATCH 29/41] btrfs: " Tejun Heo
2010-09-03 10:29 ` [PATCH 30/41] gfs2: " Tejun Heo
2010-09-03 10:29 ` [PATCH 31/41] reiserfs: " Tejun Heo
2010-09-03 10:29 ` [PATCH 32/41] nilfs2: " Tejun Heo
2010-09-03 10:29 ` [PATCH 33/41] jbd: " Tejun Heo
2010-09-03 10:29 ` [PATCH 34/41] jbd2: Modify ASYNC_COMMIT code to not rely on queue draining on barrier Tejun Heo
2010-09-03 10:29 ` [PATCH 35/41] jbd2: replace barriers with explicit flush / FUA usage Tejun Heo
2010-09-03 10:29 ` [PATCH 36/41] ext4: do not send discards as barriers Tejun Heo
2010-09-03 10:29 ` [PATCH 37/41] fat: " Tejun Heo
2010-09-03 10:29 ` [PATCH 38/41] swap: " Tejun Heo
2010-09-03 10:29 ` [PATCH 39/41] block: remove the WRITE_BARRIER flag Tejun Heo
2010-09-03 10:29 ` [PATCH 40/41] block: remove the BLKDEV_IFL_BARRIER flag Tejun Heo
2010-09-03 10:29 ` [PATCH 41/41] block: remove the BH_Eopnotsupp flag Tejun Heo
2010-09-03 18:53 ` [PATCHSET #upstream] block, fs: replace HARDBARRIER with FLUSH/FUA, take#2 Jens Axboe
2010-09-08 16:12 ` Tejun Heo
2010-09-10 10:19   ` Jens Axboe

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4C8CBFC3.8060706@oracle.com \
    --to=tao.ma@oracle.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=tj@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).