All of lore.kernel.org
 help / color / mirror / Atom feed
From: Andi Kleen <andi@firstfloor.org>
To: Mikulas Patocka <mpatocka@redhat.com>
Cc: Andi Kleen <andi-suse@firstfloor.org>,
	linux-kernel@vger.kernel.org, xfs@oss.sgi.com,
	Andi Kleen <andi@firstfloor.org>,
	Alasdair G Kergon <agk@redhat.com>, Milan Broz <mbroz@redhat.com>
Subject: Re: Device loses barrier support (was: Fixed patch for simple barriers.)
Date: Fri, 5 Dec 2008 02:37:39 +0100	[thread overview]
Message-ID: <20081205013739.GZ6703@one.firstfloor.org> (raw)
In-Reply-To: <Pine.LNX.4.64.0812041948400.14114@hs20-bc2-1.build.redhat.com>

> * barrier support in md-raid1 deviates from the specification at 
> Documentation/block/barrier.txt. The specification says that requests 
> submitted after the barrier request hit the media after the barrier 
> request hits the media. The reality is that the barrier request can be 
> randomly aborted and the requests submitted after it hit the media before 
> the barrier request.

Yes the spec should be probably updated.

But also see Linus' rant from yesterday about code vs documentation.
When in doubt the code wins.
> 
> * the filesystems developed hacks to work around this issue, the hacks 
> involve not submitting more requests after the barrier request, 

I suspect the reason the file systems did it this way is that
it was a much simpler change than to rewrite the transaction
manager for this.

> synchronously waiting for the barrier request and eventually retrying it. 
> These hacks suppress any performance advantage barriers could bring.
> 
> * you submit a patch that makes barriers even more often deviate from the 
> specification and you argue that the patch is correct because filesystems 
> handle this deviation.

Sorry what counts is the code behaviour, not the specification.

-Andi

-- 
ak@linux.intel.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

WARNING: multiple messages have this Message-ID (diff)
From: Andi Kleen <andi@firstfloor.org>
To: Mikulas Patocka <mpatocka@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>,
	linux-kernel@vger.kernel.org, xfs@oss.sgi.com,
	Alasdair G Kergon <agk@redhat.com>,
	Andi Kleen <andi-suse@firstfloor.org>,
	Milan Broz <mbroz@redhat.com>
Subject: Re: Device loses barrier support (was: Fixed patch for simple barriers.)
Date: Fri, 5 Dec 2008 02:37:39 +0100	[thread overview]
Message-ID: <20081205013739.GZ6703@one.firstfloor.org> (raw)
In-Reply-To: <Pine.LNX.4.64.0812041948400.14114@hs20-bc2-1.build.redhat.com>

> * barrier support in md-raid1 deviates from the specification at 
> Documentation/block/barrier.txt. The specification says that requests 
> submitted after the barrier request hit the media after the barrier 
> request hits the media. The reality is that the barrier request can be 
> randomly aborted and the requests submitted after it hit the media before 
> the barrier request.

Yes the spec should be probably updated.

But also see Linus' rant from yesterday about code vs documentation.
When in doubt the code wins.
> 
> * the filesystems developed hacks to work around this issue, the hacks 
> involve not submitting more requests after the barrier request, 

I suspect the reason the file systems did it this way is that
it was a much simpler change than to rewrite the transaction
manager for this.

> synchronously waiting for the barrier request and eventually retrying it. 
> These hacks suppress any performance advantage barriers could bring.
> 
> * you submit a patch that makes barriers even more often deviate from the 
> specification and you argue that the patch is correct because filesystems 
> handle this deviation.

Sorry what counts is the code behaviour, not the specification.

-Andi

-- 
ak@linux.intel.com

  reply	other threads:[~2008-12-05  1:31 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <Pine.LNX.4.64.0812040009340.15169@hs20-bc2-1.build.redhat.com>
     [not found] ` <20081204100050.GN6703@one.firstfloor.org>
2008-12-04 14:00   ` Device loses barrier support (was: Fixed patch for simple barriers.) Mikulas Patocka
2008-12-04 14:00     ` Mikulas Patocka
2008-12-04 14:20     ` Andi Kleen
2008-12-04 14:20       ` Andi Kleen
2008-12-04 14:17       ` Mikulas Patocka
2008-12-04 14:17         ` Mikulas Patocka
2008-12-04 14:58         ` Andi Kleen
2008-12-04 14:58           ` Andi Kleen
2008-12-04 16:45           ` Mikulas Patocka
2008-12-04 16:45             ` Mikulas Patocka
2008-12-04 17:48             ` Andi Kleen
2008-12-04 17:48               ` Andi Kleen
2008-12-04 17:53               ` Christoph Hellwig
2008-12-04 17:53                 ` Christoph Hellwig
2008-12-04 19:37               ` Mikulas Patocka
2008-12-04 19:37                 ` Mikulas Patocka
2008-12-04 22:15                 ` Andi Kleen
2008-12-04 22:15                   ` Andi Kleen
2008-12-04 23:08                   ` Mikulas Patocka
2008-12-04 23:08                     ` Mikulas Patocka
2008-12-05  0:48                     ` Andi Kleen
2008-12-05  0:48                       ` Andi Kleen
2008-12-05  1:16                       ` Mikulas Patocka
2008-12-05  1:16                         ` Mikulas Patocka
2008-12-05  1:37                         ` Andi Kleen [this message]
2008-12-05  1:37                           ` Andi Kleen
2008-12-05  2:21                           ` Mikulas Patocka
2008-12-05  2:21                             ` Mikulas Patocka
2008-12-05  3:09                             ` Andi Kleen
2008-12-05  3:09                               ` Andi Kleen
2008-12-05 11:52                           ` Alan Cox
2008-12-05 11:52                             ` Alan Cox
2008-12-05 12:29                             ` Andi Kleen
2008-12-05 12:29                               ` Andi Kleen
2008-12-05  3:26                 ` Device loses barrier support Eric Sandeen
2008-12-05  3:26                   ` Eric Sandeen
2008-12-07  4:17             ` Device loses barrier support (was: Fixed patch for simple barriers.) Dave Chinner
2008-12-07  4:17               ` Dave Chinner
2008-12-05  5:44           ` Device loses barrier support Timothy Shimmin
2008-12-05  5:44             ` Timothy Shimmin
     [not found] <bFnGM-7mA-5@gated-at.bofh.it>
     [not found] ` <bFpyX-1jh-15@gated-at.bofh.it>
     [not found]   ` <bFquY-2x2-5@gated-at.bofh.it>
     [not found]     ` <bFsnc-51o-37@gated-at.bofh.it>
     [not found]       ` <bFuyC-7Zt-11@gated-at.bofh.it>
     [not found]         ` <bFvuI-L8-29@gated-at.bofh.it>
     [not found]           ` <bFwTJ-2oc-13@gated-at.bofh.it>
     [not found]             ` <bFxwq-3g7-1@gated-at.bofh.it>
     [not found]               ` <bFxGa-3pz-1@gated-at.bofh.it>
     [not found]                 ` <bFHvW-7up-5@gated-at.bofh.it>
     [not found]                   ` <bFHPd-84I-13@gated-at.bofh.it>
2008-12-05 18:21                     ` Device loses barrier support (was: Fixed patch for simple barriers.) Bodo Eggert
2008-12-05 18:21                       ` Bodo Eggert
2008-12-05 18:41                       ` Andi Kleen
2008-12-05 18:41                         ` Andi Kleen

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=20081205013739.GZ6703@one.firstfloor.org \
    --to=andi@firstfloor.org \
    --cc=agk@redhat.com \
    --cc=andi-suse@firstfloor.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mbroz@redhat.com \
    --cc=mpatocka@redhat.com \
    --cc=xfs@oss.sgi.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.