Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Dave Jones]

On Thu, May 11, 2006 at 12:15:10AM -0700, Andrew Morton wrote:
 > 
 > The patch titled
 > 
 >      deprecate smbfs in favour of cifs
 > 
 > has been added to the -mm tree.  Its filename is
 > 
 >      deprecate-smbfs-in-favour-of-cifs.patch
 > 
 > See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
 > out what to do about this
 > 
 > 
 > From: Andrew Morton <akpm@osdl.org>
 > 
 > smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
 > the first five mount attempts - tell them to switch to CIFS.
 > 
 > Come November we'll mark it BROKEN and see what happens.


For Fedora Core 5, I disabled SMBFS for pretty much the same reasons.
Users migrating to CIFS haven't really had any problems so far, except for
this case: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=186914
(Which has also come up a few times on Fedora mailing lists since).

I mailed Steve about this, and he did reply, but I can't seem to find it
right now

		Dave

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

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Jesper Juhl]

On 5/11/06, akpm@osdl.org <akpm@osdl.org> wrote:
>
> The patch titled
>
>      deprecate smbfs in favour of cifs
>
> has been added to the -mm tree.  Its filename is
>
>      deprecate-smbfs-in-favour-of-cifs.patch
>
> See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
> out what to do about this
>
>
> From: Andrew Morton <akpm@osdl.org>
>
> smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
> the first five mount attempts - tell them to switch to CIFS.
>
> Come November we'll mark it BROKEN and see what happens.
>
[snip]

Perhaps an addition to  Documentation/feature-removal-schedule.txt  is
also in order?

Something noting that it will be marked as broken in November and go
away some 12 - 18 months after that perhaps?


-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Andrew Morton]

"Jesper Juhl" <jesper.juhl@gmail.com> wrote:
>
> On 5/11/06, akpm@osdl.org <akpm@osdl.org> wrote:
> >
> > The patch titled
> >
> >      deprecate smbfs in favour of cifs
> >
> > has been added to the -mm tree.  Its filename is
> >
> >      deprecate-smbfs-in-favour-of-cifs.patch
> >
> > See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
> > out what to do about this
> >
> >
> > From: Andrew Morton <akpm@osdl.org>
> >
> > smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
> > the first five mount attempts - tell them to switch to CIFS.
> >
> > Come November we'll mark it BROKEN and see what happens.
> >
> [snip]
> 
> Perhaps an addition to  Documentation/feature-removal-schedule.txt  is
> also in order?

That seems a bit duplicative, so I didn't bother.

> Something noting that it will be marked as broken in November and go
> away some 12 - 18 months after that perhaps?

We'll see.  We'd like to remove it as early as poss, of course.  But right
now, I don't know when that'll be.

The personal challenge is to remove it before Greg gets his devfs-removal
patches in ;)

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[John Kelly]

On Thu, 11 May 2006 11:27:18 -0700, Andrew Morton <akpm@osdl.org>
wrote:

>We'll see.  We'd like to remove it as early as poss, of course.  But right
>now, I don't know when that'll be.

>The personal challenge is to remove it before Greg gets his devfs-removal
>patches in ;)

I read there might be a bugfix release.  A garbage removal release
would help too.  Then it would be easier to see the bugs.

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Jan Engelhardt]

> > The patch titled
> > 
> >      deprecate smbfs in favour of cifs
> > 
> > has been added to the -mm tree.  Its filename is
> > 
> >      deprecate-smbfs-in-favour-of-cifs.patch
> > 
> > See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
> > out what to do about this
> > 
> > 
> > From: Andrew Morton <akpm@osdl.org>
> > 
> > smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
> > the first five mount attempts - tell them to switch to CIFS.
> > 
> > Come November we'll mark it BROKEN and see what happens.

Sorry for falling in late but we can't do that.
Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS. 
Furthermore, there seems to be a strange CIFS error when trying to do so 
(varying error codes)...:

13:11 shanghai:/etc # mount //wideland/hda1 /mnt/wideland -t smbfs -o password=realw
13:11 shanghai:/etc # ls /mnt/wideland
.               cygwin             DRVSPACE.BIN                      msdos.sys
..              tcpp               IO.SYS                            system.1st
DA              windows            SCANDISK.LOG                      tools.conf
Eigene Dateien  AUTOEXEC.BAT       SETUPXLG.TXT
Programme       COMMAND.COM        Verkn?pfung mit Scandisk.log.lnk
RECYCLED        CYGWIN_SYSLOG.TXT  config.sys
13:11 shanghai:/etc # umount /mnt/wideland
13:11 shanghai:/etc # mount //wideland/hda1 /mnt/wideland -t cifs -o password=realw
mount error 2 = No such file or directory
Refer to the mount.cifs(8) manual page (e.g.man mount.cifs)
13:11 shanghai:/etc # mount //wideland/hda1 /mnt/wideland -t cifs -o password=realw
mount error 112 = Host is down
Refer to the mount.cifs(8) manual page (e.g.man mount.cifs)

It's certainly not down.


Jan Engelhardt
-- 

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[John Kelly]

On Fri, 12 May 2006 17:03:56 +0200 (MEST), Jan Engelhardt
<jengelh@linux01.gwdg.de> wrote:
 
>> > smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
>> > the first five mount attempts - tell them to switch to CIFS.

>> > Come November we'll mark it BROKEN and see what happens.

>Sorry for falling in late but we can't do that.
>Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.

W98?  He's dead, Jim.

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Steven Rostedt]

On Fri, 12 May 2006, John Kelly wrote:

> On Fri, 12 May 2006 17:03:56 +0200 (MEST), Jan Engelhardt
> <jengelh@linux01.gwdg.de> wrote:
>
> >> > smbfs is a bit buggy and has no maintainer.  Change it to shout at the user on
> >> > the first five mount attempts - tell them to switch to CIFS.
>
> >> > Come November we'll mark it BROKEN and see what happens.
>
> >Sorry for falling in late but we can't do that.
> >Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.
>
> W98?  He's dead, Jim.
>

huh, my wife has a laptop that she still uses that has w98 on it. And I do
use smbfs to sometimes communicate with it.  Why upgrade when you don't
have to?

-- Steve

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[John Kelly]

On Fri, 12 May 2006 12:24:40 -0400 (EDT), Steven Rostedt
<rostedt@goodmis.org> wrote:

>> >Sorry for falling in late but we can't do that.
>> >Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.

>> W98?  He's dead, Jim.

>huh, my wife has a laptop that she still uses that has w98 on it. And I do
>use smbfs to sometimes communicate with it.

Users who need vintage features can use vintage kernels.  They haven't
been pulled off the market.


>Why upgrade when you don't have to?

Why drag around old worn out baggage in new kernels?

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[grundig]

El Fri, 12 May 2006 12:19:18 -0400,
John Kelly <jak@isp2dial.com> escribió:

> W98?  He's dead, Jim.

Agreed - Win 98 and Me support stops on July 11. If even Microsoft
stops supporting it...

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Tom Rini]

On Fri, May 12, 2006 at 12:31:02PM -0400, John Kelly wrote:
> On Fri, 12 May 2006 12:24:40 -0400 (EDT), Steven Rostedt
> <rostedt@goodmis.org> wrote:
> 
> >> >Sorry for falling in late but we can't do that.
> >> >Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.
> 
> >> W98?  He's dead, Jim.
> 
> >huh, my wife has a laptop that she still uses that has w98 on it. And I do
> >use smbfs to sometimes communicate with it.
> 
> Users who need vintage features can use vintage kernels.  They haven't
> been pulled off the market.

Having a shiny new storage box in my house that just might need to talk
with old laptops and new laptops and so on doesn't exactly jive with
that.

Of course perhaps this will cause someone who does care about smbfs to
setup up to the plate and maintain it.

-- 
Tom Rini

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Steven Rostedt]

On Fri, 12 May 2006, John Kelly wrote:

> On Fri, 12 May 2006 12:24:40 -0400 (EDT), Steven Rostedt
> <rostedt@goodmis.org> wrote:
>
> >> >Sorry for falling in late but we can't do that.
> >> >Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.
>
> >> W98?  He's dead, Jim.
>
> >huh, my wife has a laptop that she still uses that has w98 on it. And I do
> >use smbfs to sometimes communicate with it.
>
> Users who need vintage features can use vintage kernels.  They haven't
> been pulled off the market.

I need to have the latest on my desktop.  The machine I'm communicating
with is old.

>
>
> >Why upgrade when you don't have to?
>
> Why drag around old worn out baggage in new kernels?
>

Because it's still used by new kernels.

-- Steve

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Steven Rostedt]

John, on LKML it is expected to not strip CC lists and at least keep the
one you are responding to.

On Fri, 12 May 2006, Tom Rini wrote:

>
> Of course perhaps this will cause someone who does care about smbfs to
> setup up to the plate and maintain it.
>

I admit, I only communicate to that laptop about once a year.  But if need
be, I'll (shudder) try to maintain it.

-- Steve

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[John Kelly]

On Fri, 12 May 2006 09:40:34 -0700, Tom Rini
<trini@kernel.crashing.org> wrote:

>On Fri, May 12, 2006 at 12:31:02PM -0400, John Kelly wrote:
>> On Fri, 12 May 2006 12:24:40 -0400 (EDT), Steven Rostedt
>> <rostedt@goodmis.org> wrote:
 
>> >> >Sorry for falling in late but we can't do that.
>> >> >Win 98 (95 too?) shared can not be mounted with CIFS, it requires SMBFS.

>> >> W98?  He's dead, Jim.

>> >huh, my wife has a laptop that she still uses that has w98 on it. And I do
>> >use smbfs to sometimes communicate with it.

>> Users who need vintage features can use vintage kernels.  They haven't
>> been pulled off the market.

>Having a shiny new storage box in my house that just might need to talk
>with old laptops and new laptops and so on doesn't exactly jive with
>that.

If every hypothetical user has to die off before old features are
culled from the kernel, it will become a mountain of old stinking
garbage.


>Of course perhaps this will cause someone who does care about smbfs to
>setup up to the plate and maintain it.

Then let them maintain it out of tree.  People have to maintain new
features out of tree, why not old too?

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Linus Torvalds]

On Fri, 12 May 2006, John Kelly wrote:
> 
> Users who need vintage features can use vintage kernels.  They haven't
> been pulled off the market.

I disagree.

We have two cases:

 - newer kernels don't always support vintage hardware any more. We don't, 
   for example, boot on 1MB PCs (I _think_ we used to), and quite frankly, 
   if you have 4MB, I'd be surprised it worked either (and that definitely 
   used to work a long time ago).

   Similarly, we've occsionally dropped a driver just because it wasn't 
   getting maintained, and we knew it couldn't work in the state it was 
   in. So over the years, machines have stopped being supported (that 
   said, if somebody complains, we try to re-instate the driver. Most 
   dropped drivers have never even been commented upon, because they 
   really aren't used any more. When was the last time you saw an MCA 
   machine or a PC98? I bet some people on this list have never even 
   heard of either)

 - we sometimes drop sw features that have been deprecated long ago, and 
   that there are better alternatives for. That said, this is pretty damn 
   rare too. I can remember Xiafs, and devfs is obviously on that path 
   too.

But we do _not_ drop features just because they are deemed "unnecessary". 
As long as somebody actually _uses_ smbfs, and as long as those users are 
willing to test and perhaps send in patches for when/if it breaks, we 
should not drop it.

The cost of keeping a filesystem is not normally very high. The way 
filesystems in particular get deprecated is if they have really serious 
problems, and nobody ends up being able or willing to fix them at all, and 
you _can_ migrate away. But if we're talking about win98, it probably 
still actually has a pretty big user base, and most of the machines that 
run it probably really cannot upgrade.

For exactly the same reason you mention:

	"Users who need vintage features can use vintage kernels."

ie you end up having people who have vintage hardware, and they use 
vintage kernels, but in their case, the "vintage" is Win95 or Win98. That 
does't mean that the _linux_ machine they use is necessarily vintage.

		Linus

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[John Kelly]

On Fri, 12 May 2006 09:59:11 -0700 (PDT), Linus Torvalds
<torvalds@osdl.org> wrote:

>On Fri, 12 May 2006, John Kelly wrote:
 
>> Users who need vintage features can use vintage kernels.  They haven't
>> been pulled off the market.

>I disagree.

What can I say?  You're the man.  I think you maintain a great kernel,
btw.

I just think forward progress would be easier without dragging around
some of the old baggage in the kernel.

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Linus Torvalds]

On Fri, 12 May 2006, John Kelly wrote:
> 
> I just think forward progress would be easier without dragging around
> some of the old baggage in the kernel.

I think that is generally true, but we've actually been pretty successful 
in having a modular enough source tree that most of the time, old code 
simply is old - and doesn't much affect new code.

That is especially true in filesystems. We've had a few fairly painful 
times (the page cache changes in 2.3.x and the switch to the dentry cache 
in 2.1.x(?)), but on the whole we've had a pretty stable VFS interface 
that hasn't needed _that_ much work for individual filesystems.

We've had much bigger problems with drivers, although there the main 
reason for the problems is just that if some interface changes even very 
trivially, there's just so _many_ drivers that they tend to be harder to 
fix up (and they tend to do things that you can't "think about" because 
it's very much due to bugs or specific issues with some random piece of 
hardware that most developers don't even have access to).

Also, while it can be easier in _one_sense_ to move forwards if you drop 
the old stuff, it often ends up making it harder in another sense: it can 
mean, for example, that people or distributions need to do more work to 
update, which in turn can mean that you have a much harder time getting 
the change tested.

Which then in turn can mean that you actually lose more developer time 
than you gained from the code simplification..

So it's not always a very clear-cut thing. For the _users_ (and those are 
who matter most), backwards compatibility is almost always absolutely the 
biggest priority, and everything else comes second.

		Linus

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Jan Engelhardt]

>
>Why drag around old worn out baggage in new kernels?
>

Maybe merge smbfs into cifs if the protocols are not too different?



Jan Engelhardt
-- 

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Jeff V. Merkey]

Linus Torvalds wrote:

>On Fri, 12 May 2006, John Kelly wrote:
>  
>
>>Users who need vintage features can use vintage kernels.  They haven't
>>been pulled off the market.
>>    
>>
>
>I disagree.
>
>We have two cases:
>
> - newer kernels don't always support vintage hardware any more. We don't, 
>   for example, boot on 1MB PCs (I _think_ we used to), and quite frankly, 
>   if you have 4MB, I'd be surprised it worked either (and that definitely 
>   used to work a long time ago).
>
>   Similarly, we've occsionally dropped a driver just because it wasn't 
>   getting maintained, and we knew it couldn't work in the state it was 
>   in. So over the years, machines have stopped being supported (that 
>   said, if somebody complains, we try to re-instate the driver. Most 
>   dropped drivers have never even been commented upon, because they 
>   really aren't used any more. When was the last time you saw an MCA 
>   machine or a PC98? I bet some people on this list have never even 
>   heard of either)
>
> - we sometimes drop sw features that have been deprecated long ago, and 
>   that there are better alternatives for. That said, this is pretty damn 
>   rare too. I can remember Xiafs, and devfs is obviously on that path 
>   too.
>
>But we do _not_ drop features just because they are deemed "unnecessary". 
>As long as somebody actually _uses_ smbfs, and as long as those users are 
>willing to test and perhaps send in patches for when/if it breaks, we 
>should not drop it.
>
>The cost of keeping a filesystem is not normally very high. The way 
>filesystems in particular get deprecated is if they have really serious 
>problems, and nobody ends up being able or willing to fix them at all, and 
>you _can_ migrate away. But if we're talking about win98, it probably 
>still actually has a pretty big user base, and most of the machines that 
>run it probably really cannot upgrade.
>
>For exactly the same reason you mention:
>
>	"Users who need vintage features can use vintage kernels."
>
>ie you end up having people who have vintage hardware, and they use 
>vintage kernels, but in their case, the "vintage" is Win95 or Win98. That 
>does't mean that the _linux_ machine they use is necessarily vintage.
>
>		Linus
>
>  
>
Correct call. SMBFS is also very stable and well tested.

Jeff

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Jeff V. Merkey]

>
>
>>Why drag around old worn out baggage in new kernels?
>>
>>    
>>
If we follow this logic, then.

"LKML == Wikipedia"

"Your contributions can be reverted and removed at any time by any editor."

So why contribute in the first place?

Leave smbfs. The author worked his butt off and contributed to Linux 
with some very useful technology,
and leave his name in contributors.

Jeff

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Andrew Morton]

Linus Torvalds <torvalds@osdl.org> wrote:
>
> The cost of keeping a filesystem is not normally very high. The way 
>  filesystems in particular get deprecated is if they have really serious 
>  problems, and nobody ends up being able or willing to fix them at all, and 
>  you _can_ migrate away.

That's the case with smbfs and cifs, soon.

> But if we're talking about win98, it probably 
>  still actually has a pretty big user base, and most of the machines that 
>  run it probably really cannot upgrade.

cifs doesn't support w98 and w95 properly yet.  Steve's working on
it, and we hope to have that in place for 2.6.18.

So at this stage, 2.6.18 still appears to be a good time to start pushing
people toward cifs, and December looks like an appropriate time to mark
smbfs as broken.  Subject to, of course, feedback-from-the-field.

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Dave Jones]

On Sat, May 13, 2006 at 08:11:44PM -0700, Andrew Morton wrote:

 > So at this stage, 2.6.18 still appears to be a good time to start pushing
 > people toward cifs, and December looks like an appropriate time to mark
 > smbfs as broken.  Subject to, of course, feedback-from-the-field.

I'm surprised that other vendors are actually still shipping it[1].
(Not only that, some vendors have actually been sitting on smbfs
 patches for well over a year).

Given that it's clearly abandoned, moving to cifs seems to be the
only sensible thing to do, and anything that can be done to ease
that transition should be done.

		Dave

[1] Especially after the recent security problem where smbfs stayed
vulnerable for a week or so after CIFS got fixed.  How many bad guys
thought "Hmm, wonder if smbfs has the same bug" in that week?

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

Re: + deprecate-smbfs-in-favour-of-cifs.patch added to -mm tree

[Helge Hafting]

grundig wrote:

>El Fri, 12 May 2006 12:19:18 -0400,
>John Kelly <jak@isp2dial.com> escribió:
>
>  
>
>>W98?  He's dead, Jim.
>>    
>>
>
>Agreed - Win 98 and Me support stops on July 11. If even Microsoft
>stops supporting it...
>
... then people runs more linux in order to get our superior support. ;-)

Helge Hafting