Improving zvol support

[Steven Hartland]

----- Original Message ----- 
From: "Andriy Gapon" <avg at FreeBSD.org>
To: "Steven Hartland" <smh at FreeBSD.org>; <zfs-devel at FreeBSD.org>
Sent: Saturday, August 31, 2013 3:40 PM
Subject: Re: Improving zvol support


> on 30/08/2013 17:22 Steven Hartland said the following:
>> There's a few outstanding PR's for zvol issues such as:-
>> http://www.freebsd.org/cgi/query-pr.cgi?pr=178999
>> 
>> I've attached a patch to this PR (also to this email)
>> which helps clear quite a few issues highlighted by this
>> PR and by tests in STF.
>> 
>> The locking side of this patch while not 100% ideal does
>> prevent the common issues such as zvol renames so I think
>> worth getting this committing until more time can be spent
>> on it.
>> 
>> What do people think?
> 
> A few general comments:
> - it's a good idea to describe the changes and possibly to provide a patch per
> problem
> - this looks like an svn diff output, for reviewers' convenience please use -x
> -p options
> - both I and Spectra Logic have independent patches to zvol code that
> significantly change locking in it and completely remove spa_namespace_lock problem
> 
> I have never seen the Spectra's code although I did have a chance.  Will Andrews
> convinced me that their approach is superior to mine and they were willing to
> upstream their changes.  I think that it would be a good idea to take a look at
> all proposed approaches, select or fuse the best and finally commit something.
> 
> BTW, the current zvol code has another major GEOM-related bug - it drops and
> picks up g_topology_lock in zvol_geom_access, which must not be done.
> 
> P.S. message-id of my proposal is <510E9FFC.6000303 at FreeBSD.org>, it was posted
> to this list.

I think we need a concerted effort to make what everyones working on or has
worked on but hasnt committed to the tree more visible.

It seems where's a lot of duplication of effort recently, which is just
wasteing everyones time.

It seems the guys over at Spectra have loads of great fixes and features
which are taking significant effort to develop yet are unknown till others
like myself look to actually get a fix into the tree which I'm sure everyone
would agree is very frustrating.

I hence propose that everyone creates a list of features / fixes that they
have locally and we start a new thread to collate these so we don't waste
valuable time duplicating effort.

    Regards
    Steve

================================================
This e.mail is private and confidential between Multiplay (UK) Ltd. and the person or entity to whom it is addressed. In the event of misdirection, the recipient is prohibited from using, copying, printing or otherwise disseminating it or any information contained in it. 

In the event of misdirection, illegible or incomplete transmission please telephone +44 845 868 1337
or return the E.mail to postmaster at multiplay.co.uk.