mount throws a nikita-3228 on me

mount throws a nikita-3228 on me

[mjt]

Hi

I took a snapshot from the Namesys BK some days ago, compiled the
2.6.6-mm2 with high hopes of the future, but it doesn't mount.

It says only small keys are supported. I think it was a Nikita-3228.[1]

Looking at the code, it seems like it checks which size of keys are
supported. AFAIK this is not an mkfs-time option, but a kernel option only?

What does it mean, is the BK kernel broken or must I work some magick
to get it working?

What does it mean when the kernel help says it will make my system
binary-incompatible? Just that it doesn't mount or will actual ELF
files break?

Why support two key sizes at all? Small keys save disk space but aren't
as cool? Is there some practical situation where one would not want
large keys?

I'm not the only one who has experienced this, but I haven't heard
anyone getting it to work, so any help would be much appreciated.

[1]
Whoever came up with this bug naming scheme deserves an award for
innovation, practice and cleanliness. If I ever get a project where
this may be applied, I will borrow the idea :)

-- 
mjt

Re: mount throws a nikita-3228 on me

[Vladimir Saveliev]

Hello

On Tue, 2004-05-25 at 12:00, Markus TÆrnqvist wrote:
> Hi
> 
> I took a snapshot from the Namesys BK some days ago, compiled the
> 2.6.6-mm2 with high hopes of the future, but it doesn't mount.
> 
> It says only small keys are supported. I think it was a Nikita-3228.[1]
> 

You seem to get reiser4 with an error which makes reiser4 to not compile
properly.

Please try to add the below 
#include <linux/types.h>
#include <linux/fs.h>
#include <asm/hardirq.h>
#include <linux/sched.h>

to reiser4.h right after line:
#include <linux/errno.h>

and remote
/* for __u?? types */
#include <linux/types.h>
/* for struct super_block, etc */
#include <linux/fs.h>
/* for in_interrupt() */
#include <asm/hardirq.h>

#include <linux/sched.h>


from debug.h 

> Looking at the code, it seems like it checks which size of keys are
> supported. AFAIK this is not an mkfs-time option, but a kernel option only?
> 
> What does it mean, is the BK kernel broken or must I work some magick
> to get it working?
> 
> What does it mean when the kernel help says it will make my system
> binary-incompatible? Just that it doesn't mount or will actual ELF
> files break?
> 
> Why support two key sizes at all? Small keys save disk space but aren't
> as cool? Is there some practical situation where one would not want
> large keys?
> 
> I'm not the only one who has experienced this, but I haven't heard
> anyone getting it to work, so any help would be much appreciated.
> 
> [1]
> Whoever came up with this bug naming scheme deserves an award for
> innovation, practice and cleanliness. If I ever get a project where
> this may be applied, I will borrow the idea :)

Re: mount throws a nikita-3228 on me

[Alex Zarochentsev]

On Tue, May 25, 2004 at 12:30:57PM +0400, Vladimir Saveliev aka Monstr wrote:
> Hello
> 
> On Tue, 2004-05-25 at 12:00, Markus Törnqvist wrote:
> > Hi
> > 
> > I took a snapshot from the Namesys BK some days ago, compiled the
> > 2.6.6-mm2 with high hopes of the future, but it doesn't mount.
> > 
> > It says only small keys are supported. I think it was a Nikita-3228.[1]
> > 
> 
> You seem to get reiser4 with an error which makes reiser4 to not compile
> properly.

it was missing 
#include <linux/config.h> in reiser4.h

I just added it to our BK.

 
-- 
Alex.

Re: mount throws a nikita-3228 on me

[Alex Zarochentsev]

On Tue, May 25, 2004 at 12:55:35PM +0400, Alex Zarochentcev wrote:
> On Tue, May 25, 2004 at 12:30:57PM +0400, Vladimir Saveliev aka Monstr wrote:
> > Hello
> > 
> > On Tue, 2004-05-25 at 12:00, Markus Törnqvist wrote:
> > > Hi
> > > 
> > > I took a snapshot from the Namesys BK some days ago, compiled the
> > > 2.6.6-mm2 with high hopes of the future, but it doesn't mount.
> > > 
> > > It says only small keys are supported. I think it was a Nikita-3228.[1]
> > > 
> > 
> > You seem to get reiser4 with an error which makes reiser4 to not compile
> > properly.
> 
> it was missing 
> #include <linux/config.h> in reiser4.h
> 
> I just added it to our BK.

==================================================
# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
#   2004/05/24 23:59:58+04:00 zam@crimson.namesys.com 
#   include linux/config.h  explicitly into reiser4.h
# 
# reiser4.h
#   2004/05/24 23:59:54+04:00 zam@crimson.namesys.com +1 -0
#   include linux/config.h  explicitly into reiser4.h
# 
# debug.c
#   2004/05/24 23:59:54+04:00 zam@crimson.namesys.com +1 -0
#   add #include "znode.h" for print_znodes(). 
# 
diff -Nru a/debug.c b/debug.c
--- a/debug.c	Tue May 25 13:06:00 2004
+++ b/debug.c	Tue May 25 13:06:00 2004
@@ -33,6 +33,7 @@
 #include "context.h"
 #include "super.h"
 #include "txnmgr.h"
+#include "znode.h"
 
 #include <linux/sysfs.h>
 #include <linux/slab.h>
diff -Nru a/reiser4.h b/reiser4.h
--- a/reiser4.h	Tue May 25 13:06:00 2004
+++ b/reiser4.h	Tue May 25 13:06:00 2004
@@ -6,6 +6,7 @@
 #if !defined( __REISER4_H__ )
 #define __REISER4_H__
 
+#include <linux/config.h>
 #include <asm/param.h>		/* for HZ */
 #include <linux/errno.h>
 
==================================================
> -- 
> Alex.

-- 
Alex.

Re: mount throws a nikita-3228 on me

[mjt]

On Tue, May 25, 2004 at 12:30:57PM +0400, Vladimir Saveliev wrote:

>> It says only small keys are supported. I think it was a Nikita-3228.[1]
>You seem to get reiser4 with an error which makes reiser4 to not compile
>properly.

Ok...

>Please try to add the below 
>#include <linux/types.h>
>#include <linux/fs.h>
>#include <asm/hardirq.h>
>#include <linux/sched.h>
>
>to reiser4.h right after line:
>#include <linux/errno.h>
>
>and remote
>/* for __u?? types */
>#include <linux/types.h>
>/* for struct super_block, etc */
>#include <linux/fs.h>
>/* for in_interrupt() */
>#include <asm/hardirq.h>
>
>#include <linux/sched.h>
>
>
>from debug.h 

Do you mean remove?

I took a new BK pull but it still has the same problem..

I will try this one more time today, but probably not report until
tomorrow morning. I need sleep ;)

BTW, just in case some other people are wondering the same as I, some
Reiser4 patch set kernels like -love5 are ported from older Namesys
code. I thought to let people know, because I thought otherwise :)

And thanks for the help, guys!

-- 
mjt

Re: mount throws a nikita-3228 on me

[Alex Zarochentsev]

On Wed, May 26, 2004 at 12:11:33AM +0300, Markus Törnqvist wrote:
> On Tue, May 25, 2004 at 12:30:57PM +0400, Vladimir Saveliev wrote:
> 
> >> It says only small keys are supported. I think it was a Nikita-3228.[1]
> >You seem to get reiser4 with an error which makes reiser4 to not compile
> >properly.
> 
> Ok...
> 
> >Please try to add the below 
> >#include <linux/types.h>
> >#include <linux/fs.h>
> >#include <asm/hardirq.h>
> >#include <linux/sched.h>
> >
> >to reiser4.h right after line:
> >#include <linux/errno.h>
> >
> >and remote
> >/* for __u?? types */
> >#include <linux/types.h>
> >/* for struct super_block, etc */
> >#include <linux/fs.h>
> >/* for in_interrupt() */
> >#include <asm/hardirq.h>
> >
> >#include <linux/sched.h>
> >
> >
> >from debug.h 
> 
> Do you mean remove?
> 
> I took a new BK pull but it still has the same problem..

There is a delay (up to 24h) in the public BK update. 

> I will try this one more time today, but probably not report until
> tomorrow morning. I need sleep ;)
> 
> BTW, just in case some other people are wondering the same as I, some
> Reiser4 patch set kernels like -love5 are ported from older Namesys
> code. I thought to let people know, because I thought otherwise :)
> 
> And thanks for the help, guys!
> 
> -- 
> mjt
> 

-- 
Alex.

Re: mount throws a nikita-3228 on me

[mjt]

On Wed, May 26, 2004 at 09:20:53AM +0400, Alex Zarochentsev wrote:
>
>There is a delay (up to 24h) in the public BK update. 

OK. I tried it with what was suggested on the list.
It did not, however work. I added the lines to reiser4.h; should I
have seen some better debug info? I just did a quick boot, so I
may have missed something...

I understand times have been hectic at Namesys, so just keep on
trying and I'll keep on testing and telling you what's up :)

But I understood from IRC that this problem should really be fixed?
So am I doing something wrong here?

I just noted that I don't have large keys enabled in the kernel.
Must have missed that one. Anyway, still it fails and says small
keys are supported, alone.

Thanks for the help, though, and I'll get back to you

-- 
mjt

Re: mount throws a nikita-3228 on me

[Vladimir Saveliev]

Hello

On Thu, 2004-05-27 at 00:40, Markus TÆrnqvist wrote:
> On Wed, May 26, 2004 at 09:20:53AM +0400, Alex Zarochentsev wrote:
> >
> >There is a delay (up to 24h) in the public BK update. 
> 
> OK. I tried it with what was suggested on the list.
> It did not, however work. I added the lines to reiser4.h; 

did you remove lines from debug.h?

> should I
> have seen some better debug info? I just did a quick boot, so I
> may have missed something...
> 
> I understand times have been hectic at Namesys, so just keep on
> trying and I'll keep on testing and telling you what's up :)
> 
> But I understood from IRC that this problem should really be fixed?
> So am I doing something wrong here?
> 
certainly
> I just noted that I don't have large keys enabled in the kernel.
> Must have missed that one. Anyway, still it fails and says small
> keys are supported, alone.
> 
> Thanks for the help, though, and I'll get back to you

Please try the following: add something incompilable in reiser4.h right
after 
#define REISER4_LARGE_KEY (1)

on compiling you should get some files not compilable - as expected
and some files which compile fine. The last ones have wrong #include
section. The mentioned change in reiser4.h and debug.h should fix that
though

Re: mount throws a nikita-3228 on me

[mjt]

On Thu, May 27, 2004 at 11:02:47AM +0400, Vladimir Saveliev wrote:

>> OK. I tried it with what was suggested on the list.
>> It did not, however work. I added the lines to reiser4.h; 
>did you remove lines from debug.h?

Yes.

>> But I understood from IRC that this problem should really be fixed?
>> So am I doing something wrong here?
>certainly

It's seldom comforting to hear you're wrong; this is an exception :)

>Please try the following: add something incompilable in reiser4.h right
>after 
>#define REISER4_LARGE_KEY (1)
>
>on compiling you should get some files not compilable - as expected
>and some files which compile fine. The last ones have wrong #include
>section. The mentioned change in reiser4.h and debug.h should fix that
>though

I actually got the bugger to mount now. I just can't remember what
I did, because it was past midnight...
I switched to large keys, that must be one thing. So small keys are
probably broken? Should it even mount in that situation?

I got an oops and the bird and some stack trace, though. So it mounts
but goes belly-up.

I disabled the Reiser4 syscalls and it still didn't work. Then I went
to sleep.

There seems to be some funkyness with compiling as well.
Sometimes it dies with a get: command not found.

Is there any chance of having some daily snapshot diff available?
This was talked about before. It would save a shitload of trouble
if one wouldn't have to mess with bk and track pulled snapshots and
so on and so forth...

I'll post more details on the errors and some screenshots of the
stack trace when I get the opportunity and my corner of the web
back working. Hopefully today. And of course my .config and everything
else relevant.

Thanks for the help!
The first step in fixing a bug is knowing of its existence ;)

-- 
mjt