Compile probs with yesterday's auto-snapshot

Compile probs with yesterday's auto-snapshot

[David Dabbs]

Yesterday I kept having problems with an "incomplete type error" compiling
yesterday's auto-snapshot. I finally ended up changing the
CONFIG_REISER4_SYSFS define in fs/reiser4/kattr.h (I think that's the file)
from (1) to (0). 

In a separate but related question, does the reiser4 patchset add the
kernel-build configurable options to the kconfig e.g.
REISER4_STATS/DEBUG/SYSFS/NOOPT/COPY_ON_CAPTURE/SYSCALL/<others?>...? I'm
sure LARGE_KEY is there, but I can't remember which other one I saw. If it
doesn't already should LARGE_KEY prompt with a default value of "Y"?

Also the READ.ME for the prior SNAPSHOT (03/26, yes ages ago I know) had the
following notes. Since the recent 08/04 snapshot only notes that "This is
mostly bug fix snapshot" should we assume that the prior notes are resolved
or that there are still issues with the items? I've made some notes below



1. Only i386 and UML on i386 are supported.
ddabbs: Big endian appears to be supported.

2. CONFIG_REISER4_FS_SYSCALL doesn't work.

3. CONFIG_REISER4_NOOPT doesn't work.

4. reiser4 has been tested as a root file system and works in read-only
   mode, but:

       . remount is not yet supported, 
       . support in boot-loaders may be missing

ddabbs: A grub patch is provided, so that comment is likely out-of-date.

9. reiser4 can now be compiled as a module.

ddabbs: I don't think this is working yet.


c. CONFIG_REISER4_COPY_ON_CAPTURE is stabler than before.

d. readdir() over NFS has known problems.

e. mmap(2)-intensive workloads (iozone benchmark in particular) have
problems in low-memory setups.


David

Re: Compile probs with yesterday's auto-snapshot

[Adrian Ulrich]

> 1. Only i386 and UML on i386 are supported.
> ddabbs: Big endian appears to be supported.

I don't think so.

Mounting a reiser4 partiton on ppc-linux doesn't work
(mount segfaults + kernel Oops)

I created the FS on x86, so it's not the fault of mkfs.reiser4
(I didn't even compile the userland-tools on ppc-linux)


Does namesys plan to support Big endian soon?

How about porting Reiser4 to eg. Solaris?
I think people would buy licenses for solaris-reiser4 :)


bye
 

Re: Compile probs with yesterday's auto-snapshot

[Alex Zarochentsev]

On Tue, Aug 10, 2004 at 02:19:54PM -0500, David Dabbs wrote:
> 
> 
> Yesterday I kept having problems with an "incomplete type error" compiling
> yesterday's auto-snapshot. I finally ended up changing the
> CONFIG_REISER4_SYSFS define in fs/reiser4/kattr.h (I think that's the file)
> from (1) to (0). 

yes. everything fine in the BK, but latest autosnapshots were made from wrong
linux core patches.  I have re-done the autosnapshot script setup and will check
the next snapshot.

> 
> In a separate but related question, does the reiser4 patchset add the
> kernel-build configurable options to the kconfig e.g.
> REISER4_STATS/DEBUG/SYSFS/NOOPT/COPY_ON_CAPTURE/SYSCALL/<others?>...? I'm
> sure LARGE_KEY is there, but I can't remember which other one I saw. If it
> doesn't already should LARGE_KEY prompt with a default value of "Y"?
> 
> Also the READ.ME for the prior SNAPSHOT (03/26, yes ages ago I know) had the
> following notes. Since the recent 08/04 snapshot only notes that "This is
> mostly bug fix snapshot" should we assume that the prior notes are resolved
> or that there are still issues with the items? I've made some notes below
> 
> 
> 
> 1. Only i386 and UML on i386 are supported.
> ddabbs: Big endian appears to be supported.
> 
> 2. CONFIG_REISER4_FS_SYSCALL doesn't work.
> 
> 3. CONFIG_REISER4_NOOPT doesn't work.
> 
> 4. reiser4 has been tested as a root file system and works in read-only
>    mode, but:
> 
>        . remount is not yet supported, 
>        . support in boot-loaders may be missing
> 
> ddabbs: A grub patch is provided, so that comment is likely out-of-date.
> 
> 9. reiser4 can now be compiled as a module.
> 
> ddabbs: I don't think this is working yet.

hmm. I thought we made all needed symbols exportable. what is the problem?
> 
> c. CONFIG_REISER4_COPY_ON_CAPTURE is stabler than before.
> 
> d. readdir() over NFS has known problems.
> 
> e. mmap(2)-intensive workloads (iozone benchmark in particular) have
> problems in low-memory setups.
> 
> 
> David
> 
> 

-- 
Alex.

Re: Compile probs with yesterday's auto-snapshot

[Hans Reiser]

Adrian Ulrich wrote:

>
>
>How about porting Reiser4 to eg. Solaris?
>I think people would buy licenses for solaris-reiser4 :)
>
>
>bye
> 
>
>
>  
>
Probably a good idea.  Wish I had money for it.

Hans

RE: Compile probs with yesterday's auto-snapshot

[David Dabbs]

> -----Original Message-----
> From: Alex Zarochentsev [mailto:zam@namesys.com]
> Sent: Tuesday, August 10, 2004 2:41 PM
> To: David Dabbs
> Cc: 'ReiserFS List'
> Subject: Re: Compile probs with yesterday's auto-snapshot
> 
> >
> > 9. reiser4 can now be compiled as a module.
> >
> > ddabbs: I don't think this is working yet.
> 
> hmm. I thought we made all needed symbols exportable. what is the problem?
> >
> --
> Alex.
[David Dabbs] 

Not based on actually trying it. The comment is based solely on recent
(obviouslt misinformed) reiserfslist traffic I saw that mentioned that. 

dd

Re: Compile probs with yesterday's auto-snapshot

[mjt]

On Tue, Aug 10, 2004 at 02:19:54PM -0500, David Dabbs wrote:
>4. reiser4 has been tested as a root file system and works in read-only
>   mode, but:
>       . remount is not yet supported, 
>       . support in boot-loaders may be missing
>ddabbs: A grub patch is provided, so that comment is likely out-of-date.

So what am I missing as I've used Reiser4 as root for a long time?

Never had any problems, so uhh, is grub the only one left?

I'm still using lilo, bugger me, and some time ago I made a separate
/boot with ReiserFS so I can't even help with grub testing :P

-- 
mjt

Re: Compile probs with yesterday's auto-snapshot

[Adrian Ulrich]

> I'm still using lilo, bugger me, and some time ago I made a separate
> /boot with ReiserFS so I can't even help with grub testing :P

Just give grub a try with a Reiser4 enabled Boot Floppy

I uploaded stage1 and stage2:
http://blinkenlights.ch/reiser4/builds/20040810/grub20040810/


-> Create the bootfloppy
     # dd if=stage1 of=/dev/fd0 bs=512 count=1
     1+0 records in
     1+0 records out
     # dd if=stage2 of=/dev/fd0 bs=512 seek=1
     153+1 records in
     153+1 records out
     #

-> Reboot from floppy and boot:
grub> root (hd0,0)                   #HDNUM,PARTITION
grub> kernel /boot/bla root=/dev/bla #load the kernel image, you can use
                                     #<TAB>, btw;)
grub>boot                            #Boot :)

Re: Compile probs with yesterday's auto-snapshot

[Vladimir V. Saveliev]

Hello

Adrian Ulrich wrote:
>>1. Only i386 and UML on i386 are supported.
>>ddabbs: Big endian appears to be supported.
> 
> 
> I don't think so.
> 
> Mounting a reiser4 partiton on ppc-linux doesn't work
> (mount segfaults + kernel Oops)
> 
> I created the FS on x86, so it's not the fault of mkfs.reiser4
> (I didn't even compile the userland-tools on ppc-linux)
> 
> 
> Does namesys plan to support Big endian soon?
> 

Yes. The only problem is lack of hardware.

Re: Compile probs with yesterday's auto-snapshot

[Adrian Ulrich]

> Probably a good idea.  Wish I had money for it.

How much would namesys need?