New Reiser4 patch

New Reiser4 patch

[Brian Uhrain]

Hi,

I just noticed that a Reiser4 patch for kernel 2.6.14.6 was posted on 
ftp.namesys.com yesterday, and I was wondering what changes (if any) 
there are to Reiser4 itself between that patch and the 2.6.15 patch that 
was released two months ago?  Does it fix the issue with the periodic 
(temporary) system freezes during I think background flushes that was 
discussed previously on the list?  And out of curiosity, is there a 
specific technical reason why the patch is for 2.6.14.6 instead of one 
of the 2.6.15 source trees?

Sorry for the barrage of questions as my first post to this list.. :D  I 
use Reiser4 as the root filesystem on all four of my systems (two 
Athlon64 laptops and a dual-core Athlon64 desktop, both using x86-64 
kernels, and a dual processor Alpha).  And I must say, keep up the good 
work!  Except for the pausing issue with the (older) 2.6.14 and 2.6.15 
patches--which I've not even noticed on my dual-core system--and the 
mmap issue (only a problem with VMware, which will hard lock a system), 
the filesystem has worked superbly for me.  I've been using it for 
probably a year or more now as my root FS and have not had any data loss 
or corruptions.  I've even been using it on my Alpha system since I got 
it about a month ago (a CS20 with dual 833MHz EV67 processors) without 
any problems--all of the system software from glibc through Xorg have 
been built at least twice on it without troubles.

Thanks,
Brian


P.S.  If you want any more information on the VMware issue, I can see if 
it still does it on one of my home systems, and try to grab kernel 
messages via serial console.  I only encountered it when I first set up 
my dual-core system at work because I made everything Reiser4, while on 
my laptops my home directory is in a separate Reiser3 partition.  My 
workaround was to put the VMware virtual disks on their own Reiserfs 
partition.

Announcements? (was: Re: New Reiser4 patch)

[Raymond A. Meijer]

On Fri 17 Mar 2006 12:25, Brian Uhrain wrote:

> I just noticed that a Reiser4 patch for kernel 2.6.14.6 was posted on
> ftp.namesys.com yesterday, and I was wondering what changes (if any)
> there are to Reiser4 itself between that patch and the 2.6.15 patch that
> was released two months ago?

Why aren't these (major) events posted to this list by the ReiserFS crew?
I'm sure that most people here would appreciate that.

Or maybe create a read-only reiserfs-announce list for this purpose?


Thanks,

Ray

Re: Announcements?

[Joe Feise]

Raymond A. Meijer wrote on 03/17/06 02:35:

> On Fri 17 Mar 2006 12:25, Brian Uhrain wrote:
> 
>> I just noticed that a Reiser4 patch for kernel 2.6.14.6 was posted on
>> ftp.namesys.com yesterday, and I was wondering what changes (if any)
>> there are to Reiser4 itself between that patch and the 2.6.15 patch that
>> was released two months ago?
> 
> Why aren't these (major) events posted to this list by the ReiserFS crew?
> I'm sure that most people here would appreciate that.
> 
> Or maybe create a read-only reiserfs-announce list for this purpose?


+1

-Joe

Re: New Reiser4 patch

[Vladimir V. Saveliev]

Hello

On Fri, 2006-03-17 at 10:25 +0000, Brian Uhrain wrote:
> Hi,
> 
> I just noticed that a Reiser4 patch for kernel 2.6.14.6 was posted on 
> ftp.namesys.com yesterday, and I was wondering what changes (if any) 
> there are to Reiser4 itself between that patch and the 2.6.15 patch that 
> was released two months ago?  Does it fix the issue with the periodic 
> (temporary) system freezes during I think background flushes that was 
> discussed previously on the list?  And out of curiosity, is there a 
> specific technical reason why the patch is for 2.6.14.6 instead of one 
> of the 2.6.15 source trees?
> 

There are guys who are using 2.6.14.6 for their product and they wanted
to try fresh reiser4. So we took reiser4 of 2.6.16-rc6-mm1 and
downported it to older kernel.
There are no substantial changes between patch for 2.6.15 and the one
for 2.6.14.6.

The reason why reiser4 in mm kernel did not get updated recently is that
we are trying here various changes: read/write performance improving, VM
interaction simplification are the most substantial ones. And these
changes are not ready yet to got out.

> Sorry for the barrage of questions as my first post to this list.. :D  I 
> use Reiser4 as the root filesystem on all four of my systems (two 
> Athlon64 laptops and a dual-core Athlon64 desktop, both using x86-64 
> kernels, and a dual processor Alpha).  And I must say, keep up the good 
> work!  Except for the pausing issue with the (older) 2.6.14 and 2.6.15 
> patches--which I've not even noticed on my dual-core system--and the 
> mmap issue (only a problem with VMware, which will hard lock a system), 
> the filesystem has worked superbly for me.  I've been using it for 
> probably a year or more now as my root FS and have not had any data loss 
> or corruptions.  I've even been using it on my Alpha system since I got 
> it about a month ago (a CS20 with dual 833MHz EV67 processors) without 
> any problems--all of the system software from glibc through Xorg have 
> been built at least twice on it without troubles.
> 
> Thanks,
> Brian
> 
> 
> P.S.  If you want any more information on the VMware issue, I can see if 
> it still does it on one of my home systems, and try to grab kernel 
> messages via serial console. 

Yes, please do so.

>  I only encountered it when I first set up 
> my dual-core system at work because I made everything Reiser4, while on 
> my laptops my home directory is in a separate Reiser3 partition.  My 
> workaround was to put the VMware virtual disks on their own Reiserfs 
> partition.
> 

Re: New Reiser4 patch

[Lexington Luthor]

Vladimir V. Saveliev wrote:
> The reason why reiser4 in mm kernel did not get updated recently is that
> we are trying here various changes: read/write performance improving, VM
> interaction simplification are the most substantial ones. And these
> changes are not ready yet to got out.
> 

Is there a publicly accessible repository (git/bk or even cvs) where we 
can see the current progress? Have the performance problems with sync 
been fixed yet?

I am putting together some new development systems and I would love to 
try things like reiser4 on at least one of them. I will be happy to test 
it out and report any bugs.

Regards,
LL

Re: New Reiser4 patch

[Brian Uhrain]

Vladimir V. Saveliev wrote:
> On Fri, 2006-03-17 at 10:25 +0000, Brian Uhrain wrote:
>>P.S.  If you want any more information on the VMware issue, I can see if 
>>it still does it on one of my home systems, and try to grab kernel 
>>messages via serial console. 
> 
> 
> Yes, please do so.

I've tried to recreate the lockup at home without success; VMware ran 
just fine.  I'd say either the problem has been fixed (when I 
encountered the lockup I was using a 2.6.12 or 2.6.13 kernel, and am 
currently using 2.6.15--in all cases vanilla kernel.org sources with the 
reiser4 patches from ftp.namesys.com), or it's a bad interaction with 
one of the other kernel components.  I mention the latter because at 
home my system has an IDE disk without any fancy partitioning, and my 
work system--in addition to being dual-core--has two SATA drives and the 
partition in question was an LVM partition striped across the two drives.

- Brian