xfs
[Top] [All Lists]

Re: [PATCH] xfs_buf_iodone_callbacks to force shutdown & resubmit buf in

To: Shyam Kaushik <shyam@xxxxxxxxxxxxxxxxx>
Subject: Re: [PATCH] xfs_buf_iodone_callbacks to force shutdown & resubmit buf in case of permanent failure
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Thu, 28 Apr 2016 09:53:15 +1000
Cc: xfs@xxxxxxxxxxx, Alex Lyakas <alex@xxxxxxxxxxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <cbaec1d0ddb8505a9d35084291d33f37@xxxxxxxxxxxxxx>
References: <a71e1aed74b2922e8ed6fb40404899e7@xxxxxxxxxxxxxx> <20160426224612.GE26977@dastard> <cbaec1d0ddb8505a9d35084291d33f37@xxxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Wed, Apr 27, 2016 at 10:29:29AM +0530, Shyam Kaushik wrote:
> Hi Dave,
> 
> I am not sure how to do lock release in this code path. Is it possible
> that you can take over this bug/patch? Thanks.

To tell the truth, I haven't even thought about how to fix it yet.
I suspect that the callbacks will need to be run, but with a new
parameter passed to xfs_iflush_done to say "still dirty, unlock
only" so that it simply unlocks and removes each object from the
buffer callback list.

I've got other things that need my attention right now (e.g agfl
size problem), so it's going to be next week before I get to this...

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>