RFC: Schedule EFS for removal.

RFC: Schedule EFS for removal.

[Dave Jones]

EFS aparently is broken (See http://bugzilla.kernel.org/show_bug.cgi?id=1020)
has been so for some time, and attempts to fix it have been rejected.
Given there's a userspace app to access these volumes, how
about scheduling it for removal ?

Signed-off-by: Dave Jones <davej@redhat.com>

--- linux-2.6/Documentation/feature-removal-schedule.txt~	2007-05-21 10:42:24.000000000 -0400
+++ linux-2.6/Documentation/feature-removal-schedule.txt	2007-05-21 10:44:04.000000000 -0400
@@ -346,3 +346,9 @@ Who:  Tejun Heo <htejun@gmail.com>
 
 ---------------------------
 
+What: EFS
+When: 2.6.24
+Why:  Unmaintained for some time, and also doesn't work.
+      Userspace replacement 'efslook' aparently works.
+      (See http://bugzilla.kernel.org/show_bug.cgi?id=1020)
+
-- 
http://www.codemonkey.org.uk

Re: RFC: Schedule EFS for removal.

[Christoph Hellwig]

On Mon, May 21, 2007 at 10:46:18AM -0400, Dave Jones wrote:
> EFS aparently is broken (See http://bugzilla.kernel.org/show_bug.cgi?id=1020)
> has been so for some time, and attempts to fix it have been rejected.
> Given there's a userspace app to access these volumes, how
> about scheduling it for removal ?

It works perfectly fine, just not with the odd blocksizes on cdroms.
SGI Australia folks are using it for exporting their irix cdrom images
to various nfs systems for example, and I added that support long after
the bug was opened.

So clear NACK for this.

Re: RFC: Schedule EFS for removal.

[Dave Jones]

On Mon, May 21, 2007 at 04:02:42PM +0100, Christoph Hellwig wrote:
 > On Mon, May 21, 2007 at 10:46:18AM -0400, Dave Jones wrote:
 > > EFS aparently is broken (See http://bugzilla.kernel.org/show_bug.cgi?id=1020)
 > > has been so for some time, and attempts to fix it have been rejected.
 > > Given there's a userspace app to access these volumes, how
 > > about scheduling it for removal ?
 > 
 > It works perfectly fine, just not with the odd blocksizes on cdroms.
 > SGI Australia folks are using it for exporting their irix cdrom images
 > to various nfs systems for example, and I added that support long after
 > the bug was opened.
 > 
 > So clear NACK for this.

Ah, that woke people up :-)
Ok, patch rescinded.  Any opinion on what to do about that bug?

"use loopback" would be one option I guess.
Given its been an open bug for four years, there's obviously
not that much interest in fixing the filesystem.

	Dave

-- 
http://www.codemonkey.org.uk

Re: RFC: Schedule EFS for removal.

[Christoph Hellwig]

> Ok, patch rescinded.  Any opinion on what to do about that bug?
> 
> "use loopback" would be one option I guess.
> Given its been an open bug for four years, there's obviously
> not that much interest in fixing the filesystem.

I don't think it's a bug, it's a lacking feature.  The error message
is quite accurate aswell.

Re: RFC: Schedule EFS for removal.

[Dave Jones]

On Mon, May 21, 2007 at 04:12:09PM +0100, Christoph Hellwig wrote:
 > > Ok, patch rescinded.  Any opinion on what to do about that bug?
 > > 
 > > "use loopback" would be one option I guess.
 > > Given its been an open bug for four years, there's obviously
 > > not that much interest in fixing the filesystem.
 > 
 > I don't think it's a bug, it's a lacking feature.  The error message
 > is quite accurate aswell.

It seems to be a regression in 2.6.x though.  From the report..

"Under 2.4.21, it doesnt do this."

	Dave

-- 
http://www.codemonkey.org.uk

Re: RFC: Schedule EFS for removal.

[Christoph Hellwig]

On Mon, May 21, 2007 at 11:20:50AM -0400, Dave Jones wrote:
> On Mon, May 21, 2007 at 04:12:09PM +0100, Christoph Hellwig wrote:
>  > > Ok, patch rescinded.  Any opinion on what to do about that bug?
>  > > 
>  > > "use loopback" would be one option I guess.
>  > > Given its been an open bug for four years, there's obviously
>  > > not that much interest in fixing the filesystem.
>  > 
>  > I don't think it's a bug, it's a lacking feature.  The error message
>  > is quite accurate aswell.
> 
> It seems to be a regression in 2.6.x though.  From the report..
> 
> "Under 2.4.21, it doesnt do this."

Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
be very surprised if 2.4.21 works.

Re: RFC: Schedule EFS for removal.

[Dave Jones]

On Mon, May 21, 2007 at 04:46:13PM +0100, Christoph Hellwig wrote:
 > On Mon, May 21, 2007 at 11:20:50AM -0400, Dave Jones wrote:
 > > On Mon, May 21, 2007 at 04:12:09PM +0100, Christoph Hellwig wrote:
 > >  > > Ok, patch rescinded.  Any opinion on what to do about that bug?
 > >  > > 
 > >  > > "use loopback" would be one option I guess.
 > >  > > Given its been an open bug for four years, there's obviously
 > >  > > not that much interest in fixing the filesystem.
 > >  > 
 > >  > I don't think it's a bug, it's a lacking feature.  The error message
 > >  > is quite accurate aswell.
 > > 
 > > It seems to be a regression in 2.6.x though.  From the report..
 > > 
 > > "Under 2.4.21, it doesnt do this."
 > 
 > Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
 > be very surprised if 2.4.21 works.

good point.  I don't see anything obvious in 2.4.22 that would have
caused a regression, so this sounds suspect.
I'll close the bugzilla out based on your comments, thanks for looking at it.

	Dave

-- 
http://www.codemonkey.org.uk

Re: RFC: Schedule EFS for removal.

[Christoph Hellwig]

On Mon, May 21, 2007 at 01:11:09PM -0400, Dave Jones wrote:
>  > Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
>  > be very surprised if 2.4.21 works.
> 
> good point.  I don't see anything obvious in 2.4.22 that would have
> caused a regression, so this sounds suspect.
> I'll close the bugzilla out based on your comments, thanks for looking at it.

Btw, I'd like to take a look at Randy's patches for bigger block sizes,
it might be quite easy to fix the remaining bits.  Does anyone have
a copy since the link doesn't work anymore?

Re: RFC: Schedule EFS for removal.

[Randy Dunlap]

On Mon, 21 May 2007 18:12:42 +0100 Christoph Hellwig wrote:

> On Mon, May 21, 2007 at 01:11:09PM -0400, Dave Jones wrote:
> >  > Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
> >  > be very surprised if 2.4.21 works.
> > 
> > good point.  I don't see anything obvious in 2.4.22 that would have
> > caused a regression, so this sounds suspect.
> > I'll close the bugzilla out based on your comments, thanks for looking at it.
> 
> Btw, I'd like to take a look at Randy's patches for bigger block sizes,
> it might be quite easy to fix the remaining bits.  Does anyone have
> a copy since the link doesn't work anymore?

Sorry about that.  The patches are back in this dir:
  http://www.xenotime.net/linux/efs/

but I don't recall much about them.

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***

Re: RFC: Schedule EFS for removal.

[Willy Tarreau]

On Mon, May 21, 2007 at 01:11:09PM -0400, Dave Jones wrote:
> On Mon, May 21, 2007 at 04:46:13PM +0100, Christoph Hellwig wrote:
>  > On Mon, May 21, 2007 at 11:20:50AM -0400, Dave Jones wrote:
>  > > On Mon, May 21, 2007 at 04:12:09PM +0100, Christoph Hellwig wrote:
>  > >  > > Ok, patch rescinded.  Any opinion on what to do about that bug?
>  > >  > > 
>  > >  > > "use loopback" would be one option I guess.
>  > >  > > Given its been an open bug for four years, there's obviously
>  > >  > > not that much interest in fixing the filesystem.
>  > >  > 
>  > >  > I don't think it's a bug, it's a lacking feature.  The error message
>  > >  > is quite accurate aswell.
>  > > 
>  > > It seems to be a regression in 2.6.x though.  From the report..
>  > > 
>  > > "Under 2.4.21, it doesnt do this."
>  > 
>  > Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
>  > be very surprised if 2.4.21 works.
> 
> good point.  I don't see anything obvious in 2.4.22 that would have
> caused a regression, so this sounds suspect.
> I'll close the bugzilla out based on your comments, thanks for looking at it.

In my experience, often when people speak about 2.4.21 (which is quite old),
they in fact refer to an RHEL3 kernel, which has a "few" addon patches :-)
While I don't think that anything related to EFS might be changed in RHEL,
may it be possible that it works as a side effect of a patch in this
kernel ?

That's just pure guess anyway.

Cheers,
Willy

Re: RFC: Schedule EFS for removal.

[Dave Jones]

On Tue, May 22, 2007 at 11:01:54PM +0200, Willy Tarreau wrote:

 > >  > > "Under 2.4.21, it doesnt do this."
 > >  > 
 > >  > Randy mentioned in bugzilla that 2.4.22 didn't work either, so I'd
 > >  > be very surprised if 2.4.21 works.
 > > 
 > > good point.  I don't see anything obvious in 2.4.22 that would have
 > > caused a regression, so this sounds suspect.
 > > I'll close the bugzilla out based on your comments, thanks for looking at it.
 > 
 > In my experience, often when people speak about 2.4.21 (which is quite old),
 > they in fact refer to an RHEL3 kernel, which has a "few" addon patches :-)
 > While I don't think that anything related to EFS might be changed in RHEL,
 > may it be possible that it works as a side effect of a patch in this
 > kernel ?
 > 
 > That's just pure guess anyway.

not unless it's a recompiled version.  As shipped, RHEL3 has..

# CONFIG_EFS_FS is not set

	Dave

-- 
http://www.codemonkey.org.uk