ent:sdXnn going to D state with latest reiser4 patch

ent:sdXnn going to D state with latest reiser4 patch

[Damien Wyart]

Hello,

Doing some tests with latest reiser4 patches from 2.6.14-mm1, I noticed
that the load of the machine never goes under 2. It comes from two
processes related to reiser4 and going to D state immediately after the
filesystems are mounted. As there are two reiser4 partitions, I get two
such processes, and the load is always higher than 2. These processes
appear as ent:sdb7 and ent:sdb11 (correspnding fs are on sdb7 and sdb11)
and same with square brackets sometimes.

I can reproduce this with 2.6.14-rc1-git4 and 2.6.13.2, with the latest
patch. I do not have this problem with previous reiser4 patch
(2.6.13-mm3 + latest reiser4-only.1.patch from ftp.namesys.com.

I guess people doing simple tests of this patch will see this behaviour,
and I am surprised nobody seems to have posted about this on the list.


Best regards,

-- 
Damien Wyart

Re: ent:sdXnn going to D state with latest reiser4 patch

[Laurent Riffard]

[-- Attachment #1: Type: text/plain, Size: 1519 bytes --]

Le 18.09.2005 22:29, Damien Wyart a écrit :
> Hello,
> 
> Doing some tests with latest reiser4 patches from 2.6.14-mm1, I noticed
> that the load of the machine never goes under 2. It comes from two
> processes related to reiser4 and going to D state immediately after the
> filesystems are mounted. As there are two reiser4 partitions, I get two
> such processes, and the load is always higher than 2. These processes
> appear as ent:sdb7 and ent:sdb11 (correspnding fs are on sdb7 and sdb11)
> and same with square brackets sometimes.
> 
> I can reproduce this with 2.6.14-rc1-git4 and 2.6.13.2, with the latest
> patch. I do not have this problem with previous reiser4 patch
> (2.6.13-mm3 + latest reiser4-only.1.patch from ftp.namesys.com.
> 
> I guess people doing simple tests of this patch will see this behaviour,
> and I am surprised nobody seems to have posted about this on the list.
> 

Hello Damien, 

I didn't notice an extra load here (Linux version 2.6.14-rc1-mm2 (laurent@antares.localdomain) (gcc version 4.0.1 (4.0.1-5mdk for Mandriva Linux release 2006.0)) #130 Sat Sep 17 10:26:06 CEST 2005). But, I can confirm that "ent:hda8" is always in D state.

[notice the *mm2* ? I'm using http://www.zip.com.au/~akpm/linux/patches/stuff/2.6.14-rc1-mm1.5.gz which is kinda rc1-mm1 without big input/sysfs changes]

Furthermore reiser4 prevents the box from suspending to disk. Found in log while suspending :
Restarting tasks...<6> Strange, ent:hda8. not stopped

~~
laurent


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]