On Mon, Nov 20, 2006 at 11:45:56PM +0100, Peter Falth wrote:
> I saw that my answer did not reach the list, so here it is again
> 
> Herbert Poetzl wrote:
> >On Wed, Nov 15, 2006 at 01:54:48AM +0100, Peter Falth wrote:
> >>I get an unaligned access kernel warning when the host boots up.
> >>This is also repeated when I do vserver-stat.
> >>I have tested both Debian precompiled and compiled
> >>myself from the kernel sources with the same result.
> >
> >means you got vanilla 2.6.18.2 from kernel.org and
> >the vs2.1.1 patch and it gave the unaligned access too?
> >
> Yes a vanilla 2.6.18.2 + vs2.1.1
> also vs 2.0.2.2-rc5 gave the same unaligned access

guess we found the cause for the unaligned access
in the dietlibc readdir() function ... if you need/want
details, please contact Daniel Hokka Zakrisson

(guess it should appear sooner or later in dietlibc)

HTH,
Herbert

> >>Except for this it works great. It was my first time 
> >>using vserver
> >
> >welcome to the club!
> >
> >>Debian unstable sparc64 on Ultra 5 and 10.
> >>The unpatched kernel do not show the access errors
> >
> >unpatched means, unpatched debian or unpatched
> >vanailla from kernel.org?
> yes vanilla from kernel.org
> >
> >>Here is a copy of the error mesages
> >> 
> >>
> >>U10:/home/peter# vserver-stat  
> >>CTX   PROC    VSZ    RSS  userTIME   sysTIME    UPTIME NAME  
> >>0       69 354.5M 111.5M   2m37s41   1m00s66   1d14h25 root server  
> >>49152   19 254.7M 107.3M   1m31s90   0m14s64   1d14h22 vs1  
> >>  
> >>
> >>From kernel messages  
> >>
> >>Nov 14 14:17:29 U10 kernel: Kernel unaligned access at TPC[4bf060] 
> >>filldir64+0x68/0x140  
> >>Nov 14 14:17:29 U10 kernel: Kernel unaligned access at TPC[4bf07c] 
> >>filldir64+0x84/0x140  
> >>Nov 14 14:17:29 U10 kernel: Kernel unaligned access at TPC[4bf034] 
> >>filldir64+0x3c/0x140  
> >>Nov 14 14:17:29 U10 kernel: Kernel unaligned access at TPC[4bf060] 
> >>filldir64+0x68/0x140  
> >>Nov 14 14:17:29 U10 kernel: Kernel unaligned access at TPC[4bf07c] 
> >>filldir64+0x84/0x140
> >
> >would you be willing to run a test patch on that machine?
> >I would be interesting in investigating this, whenever you 
> >find some time  ...
> >
> No problem, send the patch and I will try it out
> 
> Peter
> 
> >TIA,
> >Herbert
> >
> >>Peter
> >>
> >>_______________________________________________
> >>Vserver mailing list
> >>[email protected]
> >>http://list.linux-vserver.org/mailman/listinfo/vserver
> 
> _______________________________________________
> Vserver mailing list
> [email protected]
> http://list.linux-vserver.org/mailman/listinfo/vserver
_______________________________________________
Vserver mailing list
[email protected]
http://list.linux-vserver.org/mailman/listinfo/vserver

Reply via email to