Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 20 Aug 2010 09:16:47 -0400
From:      jhell <jhell@DataIX.net>
To:        walter@pelissero.de
Cc:        freebsd-bugs@freebsd.org, bug-followup@freebsd.org
Subject:   Re: kern/149762: volume labels with rogue characters
Message-ID:  <4C6E803F.9030602@DataIX.net>
In-Reply-To: <19566.20916.868818.924898@zaphod.home.lan>
References:  <4C6C7518.7010906@DataIX.net> <19565.8946.444567.257102@zaphod.home.lan> <4C6D5D6F.5060907@DataIX.net> <19566.20916.868818.924898@zaphod.home.lan>

next in thread | previous in thread | raw e-mail | index | archive | help
On 08/20/2010 05:58, Walter C. Pelissero wrote:
> jhell writes:
>  > Putting aside whats written below, please don't feel that this is an
>  > attack on your work to make this happen.
> 
> Attack on my work?
> Amusing.
> 
> 
>  > I feel your work is very viable for what it is supposed to do but
>  > to the same effect does not address the issue that this might
>  > happen elsewhere to anything new that may be introduced to the
>  > system at a later point in time
> 
> What don't you understand in:
>   Now kern.geom.label.sanitation can be set to zero (default)
>   obtaining the usual behaviour without "hack".

I understand that completely still it is not a proper fix for the situation.

> 
> 
>  > In a previous post I mentioned a proper full term solution based on
>  > adjusting devfs/devd instead that would fit future situations and
>  > not call for individual adjustments for pieces of geom or anything
>  > related.
> 
> I failed to see the devfs/devd solution you mention.

I believe I meant patch devfs/devd here rather than adjust. Since were
dealing with what shows up in the device file-system why not patch that
instead of horrifying pieces of the system, one at a time.

vfs.devfs.``normalization'' - Would probably suit this. I don't expect
you to understand this... ;)

> 
> 
>  > It has be ~6 years that GEOM_LABEL was introduced to the system.
>  > r131476 | pjd | 2004-07-02 15:40:36 -0400 (Fri, 02 Jul 2004) | 22 lines
> 
> conf/37569
> Mon Apr 29 08:40:01 PDT 2002
> 
> The common issue of all the already mentioned PRs is: mounting volumes
> with names that are not compatible with fstab (and a lot of other
> things).  Which this modification of geom_label could mitigate.
> 
> There is a related PR of mine, which tackles the problem from the
> fstab stand point, that does more or less what the other PR has been
> already suggesting since 2002 and Linux has been doing for more than a
> decade.
> 
> It was closed by a sulky baby boy on the same grounds as this.
> 

Personally I have seen enough of the hash and bash coming from anyone
here. Ive tried to make sure that what I have wrote back as a reply to
your work was neither demeaning or intrusive but yet you continue to
take the same negative standpoint with objection toward further ideas
all the while holding an immature context with a submitter.


Besides this! It is not really needed to adjust your personal system for
the likes of some labels that may contain weird characters that conflict
with how the system parses fstab(5).

Simply by using the 'link' action in devfs.conf(5) you can create a link
to a device auto-magically when the device is plugged in that would be
available for use in fstab(5).

link    label/my?edgecase myedgecase2
link	msdosfs/your?edgecase myedgecase3

This is only performed if the device appears.

In fstab(5)
/dev/myedgecase2 /headcase2 msdosfs rw,noauto 0 0
/dev/myedgecase3 /headcase3 msdosfs rw,noauto 0 0


Of course you can also adjust devd.conf(5) for example to perform
specific actions upon device entry but Ill let you figure that out.


Regards,

-- 

 jhell,v



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C6E803F.9030602>