reiser4 on 2.6.13-rc6-realtime-preempt

reiser4 on 2.6.13-rc6-realtime-preempt

[gimpel]

[-- Attachment #1: Type: text/plain, Size: 1205 bytes --]

Hello!

I'm trying to compile a recent 2.6.13-rc6 with realtime-preempt-2.6.13-
rc4-RT-V0.7.53-02 and want to add reiser4. 2.6.12-rt hardlocked very
often here.
I managed to fix up the DEFINE_SPINLOCK and compat_semaphore wait;
changes by taking a look on the changes between reiser4 for 2.6.12
and 2.6.12-rt and it first seems to compile fine but ends up with:

----8<------

  CC      init/version.o
  LD      init/built-in.o
  LD      .tmp_vmlinux1
fs/built-in.o(.text+0x92999): In function `reiser4_put_super':
: undefined reference to `rcu_barrier'
fs/built-in.o(.text+0x929a9): In function `reiser4_put_super':
: undefined reference to `rcu_barrier'
fs/built-in.o(.text+0x97989): In function `_done_sinfo':
: undefined reference to `rcu_barrier'
fs/built-in.o(.text+0x97d76): In function `_done_formatted_fake':
: undefined reference to `rcu_barrier'
fs/built-in.o(.text+0x97d86): In function `_done_formatted_fake':
: undefined reference to `rcu_barrier'
make: *** [.tmp_vmlinux1] Fehler 1

----->8-----

Maybe some has managed to fix that or knows how to do so and is willing
to help me. My C knowledge is still too limited...

regards and thanks in advance
tom


[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[Vladimir V. Saveliev]

Hello

gimpel wrote:
> Hello!
> 
> I'm trying to compile a recent 2.6.13-rc6 with realtime-preempt-2.6.13-
> rc4-RT-V0.7.53-02 and want to add reiser4. 2.6.12-rt hardlocked very
> often here.

Which patches did you use?

> I managed to fix up the DEFINE_SPINLOCK and compat_semaphore wait;
> changes by taking a look on the changes between reiser4 for 2.6.12
> and 2.6.12-rt and it first seems to compile fine but ends up with:
> 
> ----8<------
> 
>   CC      init/version.o
>   LD      init/built-in.o
>   LD      .tmp_vmlinux1
> fs/built-in.o(.text+0x92999): In function `reiser4_put_super':
> : undefined reference to `rcu_barrier'

reiser4 depends on several core kernel patches. I think you did not apply them.
You might want to try to apply this patch
ftp://ftp.namesys.com/pub/reiser4-for-2.6/2.6.12/reiser4-for-2.6.12-realtime-preempt-2.6.12-final-V0.7.51-29.patch.gz
It contains all necessary patches already.

> fs/built-in.o(.text+0x929a9): In function `reiser4_put_super':
> : undefined reference to `rcu_barrier'
> fs/built-in.o(.text+0x97989): In function `_done_sinfo':
> : undefined reference to `rcu_barrier'
> fs/built-in.o(.text+0x97d76): In function `_done_formatted_fake':
> : undefined reference to `rcu_barrier'
> fs/built-in.o(.text+0x97d86): In function `_done_formatted_fake':
> : undefined reference to `rcu_barrier'
> make: *** [.tmp_vmlinux1] Fehler 1
> 
> ----->8-----
> 
> Maybe some has managed to fix that or knows how to do so and is willing
> to help me. My C knowledge is still too limited...
> 
> regards and thanks in advance
> tom
> 

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[gimpel]

[-- Attachment #1: Type: text/plain, Size: 2834 bytes --]

On Fri, 12 Aug 2005 13:05:33 +0400
"Vladimir V. Saveliev" <vs@namesys.com> wrote:

> Hello
> 
> gimpel wrote:
> > Hello!
> > 
> > I'm trying to compile a recent 2.6.13-rc6 with realtime-
> > preempt-2.6.13- rc4-RT-V0.7.53-02 and want to add reiser4. 2.6.12-
> > rt hardlocked very often here.
> 
> Which patches did you use?
> 
I used the patches from 2.6.13-rc5-mm1 and added the changes that have
been done to reiser4-for-2.6.12-realtime-preempt-2.6.12-final-
V0.7.51-29.patch.gz which look like that: 
http://gimpel.ath.cx/~tom/r4_to_rtp.patch 
The complete patches and changes are in this diff:
http://gimpel.ath.cx/~tom/r4_for_realtime-preempt-2.6.13-rc4-RT-V0.7.53-02_notworking.patch
Seems like there where some new changes to realtime-preempt that break
reiser4 again. Maybe the is to wait for stable 2.6.13 before doing
tests with realtime-preempt as it gets updated twice a day. 
And i so much hope the kernel guys decide to merge reiser4. 

> > I managed to fix up the DEFINE_SPINLOCK and compat_semaphore wait;
> > changes by taking a look on the changes between reiser4 for 2.6.12
> > and 2.6.12-rt and it first seems to compile fine but ends up with:
> > 
> > ----8<------
> > 
> >   CC      init/version.o
> >   LD      init/built-in.o
> >   LD      .tmp_vmlinux1
> > fs/built-in.o(.text+0x92999): In function `reiser4_put_super':
> > : undefined reference to `rcu_barrier'
> 
> reiser4 depends on several core kernel patches. I think you did not
> apply them. You might want to try to apply this patch
> ftp://ftp.namesys.com/pub/reiser4-for-2.6/2.6.12/reiser4-for-2.6.12-realtime-preempt-2.6.12-final-V0.7.51-29.patch.gz
> It contains all necessary patches already.

That compiled fine with 2.6.12-RT-xx, but 2.6.12-RT didn't work very
well here. it hardlocked and performance whise there was not much
difference to vanilla. 2.6.11-RT + reiser4 works fine, but to be able to
get some sound out of my second soundcard i need a recent alsa-cvs,
which doesn't compile with kernel < 2.6.13-rcX. 
The above reiser4 patch is broken with 2.6.13-rc4-RT again.

> 
> > fs/built-in.o(.text+0x929a9): In function `reiser4_put_super':
> > : undefined reference to `rcu_barrier'
> > fs/built-in.o(.text+0x97989): In function `_done_sinfo':
> > : undefined reference to `rcu_barrier'
> > fs/built-in.o(.text+0x97d76): In function `_done_formatted_fake':
> > : undefined reference to `rcu_barrier'
> > fs/built-in.o(.text+0x97d86): In function `_done_formatted_fake':
> > : undefined reference to `rcu_barrier'
> > make: *** [.tmp_vmlinux1] Fehler 1
> > 
> > ----->8-----
> > 
> > Maybe some has managed to fix that or knows how to do so and is
> > willing to help me. My C knowledge is still too limited...
> > 
> > regards and thanks in advance
> > tom
> > 
> 

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 465 bytes --]

On Fri, 12 Aug 2005 12:09:03 +0200, gimpel said:

> reiser4 again. Maybe the is to wait for stable 2.6.13 before doing
> tests with realtime-preempt as it gets updated twice a day.
> And i so much hope the kernel guys decide to merge reiser4.

Well, reiser4 can't possibly make it into 2.6.13, as we're at -rc6 already
and Linus asked for a "quiet down" several -rc ago.  What happens when the
tree opens for 2.6.14 is a different question that I can't answer....


[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[David Masover]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Vladimir V. Saveliev wrote:
> Hello

> reiser4 depends on several core kernel patches. I think you did not
> apply them.

What kind of symptoms would not applying them cause?  I grab patches
from -mm, but only the ones with "reiser4" in the name.  They seem to
work with minimal tweaking on my part, and no one said anything when I
mentioned this before.  They do seem solid...

> You might want to try to apply this patch
> ftp://ftp.namesys.com/pub/reiser4-for-2.6/2.6.12/reiser4-for-2.6.12-realtime-preempt-2.6.12-final-V0.7.51-29.patch.gz

Should I get patches from there, instead of -mm, in the future?  Which
is more likely to be current for a given *stable* kernel,
reiser4-for-2.6 or -mm?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
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=PcmR
-----END PGP SIGNATURE-----

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[gimpel]

[-- Attachment #1: Type: text/plain, Size: 1606 bytes --]

On Fri, 12 Aug 2005 13:55:19 -0500
David Masover <ninja@slaphack.com> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Vladimir V. Saveliev wrote:
> > Hello
> 
> > reiser4 depends on several core kernel patches. I think you did not
> > apply them.
> 
> What kind of symptoms would not applying them cause?  I grab patches
> from -mm, but only the ones with "reiser4" in the name.  They seem to
> work with minimal tweaking on my part, and no one said anything when I
> mentioned this before.  They do seem solid...
> 
I think Vladimir is talking about the patches to the new inode
behaviour. I'm quite sure i checked the 2.6.13-rc4-mm patchlist about
the move-fs-to-new-behaviour-* patches etc. but maybe i should re-check
that. Currently I'm using 2.6.13-rc6-ck1 + reiser4 from 2.6.13-rc5-
mm1 and it works very well. So did -rc5 and -rc4. just wanted to give
new RTP patches a try.

But i don't want to steal the time of any devs here. It's more
important to get reiser4 merged to vanilla than spending time for a guy
who plays around with -rc kernels and -mm broken-out patches without
even the basic knowledge of C code to understand such compile errors.
 All i know that it maybe needs a 'select foo' somewhere or something.
It misses something...

But I'll just wait for stable 2.6.13 + RT and hopefully a reiser4
patch to it (or even r4 in vanilla? :)). If someone had known a quick
fix I would have been happy to continue testing around with RT etc, but
it's not worth to steal the time of the devs here :) sorry for bugging!

regards!
tom




[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: reiser4 on 2.6.13-rc6-realtime-preempt

[Vladimir V. Saveliev]

Hello

David Masover wrote:
> Vladimir V. Saveliev wrote:
> 
>>>Hello
> 
> 
>>>reiser4 depends on several core kernel patches. I think you did not
>>>apply them.
> 
> 
> What kind of symptoms would not applying them cause?  

I assumed that Gimpel did not apply core patches because he had compiling error:
fs/built-in.o(.text+0x92999): In function `reiser4_put_super': undefined reference to `rcu_barrier'.

However, the problem is that some changes are to be done to get reiser4 to work in new realtime/preempt kernels.

> I grab patches
> from -mm, but only the ones with "reiser4" in the name.  They seem to
> work with minimal tweaking on my part, and no one said anything when I
> mentioned this before.  They do seem solid...
> 

You go right way.

> 
>>>You might want to try to apply this patch
>>>ftp://ftp.namesys.com/pub/reiser4-for-2.6/2.6.12/reiser4-for-2.6.12-realtime-preempt-2.6.12-final-V0.7.51-29.patch.gz
> 
> 
> Should I get patches from there, instead of -mm, in the future?  Which
> is more likely to be current for a given *stable* kernel,
> reiser4-for-2.6 or -mm?

reiser4-for-2.6 contains code which did not get enough stability yet.

>